Gaps and challenges for addressing security threats in urban

Transcription

Deliverable D2.1
Gaps and challenges for
addressing security threats in
urban environments
Editors A. Belesiotis (ATH), D. Skoutas (ATH)
Contributors N. Bakalos (ICCS), A. Belesiotis (ATH), J. Hellriegel
(FRAUNHOFER), F. Fuchs-Kittowski (FRAUNHOFER),
A. Litke (INF), N. Papadakis (INF), N. Papadakis (SPH),
S. Pfennigschmidt (FRAUNHOFER), S. Ruscher (SYN),
E. Sauli (SYN), D. Skoutas (ATH)
Version 1.4
Date September 28, 2015
Distribution PUBLIC (PU)
City.Risks
Deliverable D2.1
Executive Summary
This document surveys the state-of-the-art in the areas related to the City.Risks
project. Our analysis starts by reviewing relevant complete and integrated solutions
in the urban security landscape. A total of 22 projects and 46 software and hardware
solutions have been analyzed. Most of these solutions constitute results of
completed and ongoing research projects. The survey focuses on the use cases,
target audience, solution types, software platforms and data sources employed in
the analyzed solutions.
Then, we look into individual areas of research that are of relevance to City.Risks.
This analysis focuses on highlighting the current advances and identifying relevant
research that can be exploited or extended within the scope of the project. For every
research area of focus, we outline the most important recent advances in relation to
the City.Risks tasks. Furthermore, based on identified gaps and challenges in the
state-of-the-art, we present research directions that can be exploited within the
scope of the project.
The first relevant area of research we review is Emergency Response and Risk
Management. More, specifically, we present the recent advances in emergency
alerting and response, augmented reality, and risks management for decision
support within the operation center. Then, we outline the most important work
related to Data Management that is of relevance to City.Risks. We focus on data
acquisition and mining, query processing, privacy and anonymization, data analytics
and route planning. Our analysis continues with the areas of Mobile Sensors and
Sensor Communication. We overview recent advances on sensor technologies and
communications, we present current work in ground monitoring and theft detection,
and we discuss relevant approaches for theft detection. Finally, we outline the stateof-the-art in software development methodologies, platform architectures, design
process models and theories, and we overview platform architectures for emergency
management systems.
This report is the first deliverable of WP2. The results of this study will serve as the
foundation for the research and development tasks in WPs 3, 4 and 5.
© City.Risks Consortium
2
City.Risks
Deliverable D2.1
Table of Contents
1. INTRODUCTION ............................................................................................. 6
2. URBAN SECURITY LANDSCAPE .......................................................................... 8
2.1. Gaps and Challenges ........................................................................................... 8
2.1.1. Use cases................................................................................................................. 8
2.1.2. Target audience ...................................................................................................... 8
2.1.3. Solution type ........................................................................................................... 9
2.1.4. Platforms................................................................................................................. 9
2.1.5. Data sources ........................................................................................................... 9
2.1.6. Features provided ................................................................................................. 10
2.1.7. Dedication to standards........................................................................................ 10
2.1.8. Gaps and challenges identified ............................................................................. 11
2.2. Solutions Overview ........................................................................................... 11
2.2.1. Research fields ...................................................................................................... 11
2.2.2. Rolled out applications ......................................................................................... 12
2.2.3. Ongoing projects’ development ........................................................................... 14
2.2.4. Solutions in focus .................................................................................................. 16
3. EMERGENCY RESPONSE AND RISKS MANAGEMENT.............................................. 27
3.1. Emergency Alerting and Response ................................................................... 27
3.1.1. Alerting ................................................................................................................. 27
3.1.2. Response ............................................................................................................... 30
3.1.3. Research directions .............................................................................................. 31
3.2. Mobile Augmented Reality (mAR) .................................................................... 31
3.2.1. User interaction with mAR data ........................................................................... 32
3.2.2. mAR content models ............................................................................................ 32
3.2.3. Generation, provision and rendering of mAR data .............................................. 33
3.3. Risks Management and Operation Center ....................................................... 34
3.3.1. Risk management ................................................................................................. 34
3.3.2. Operation center .................................................................................................. 35
3.3.3. Simulation and visualization of crime and risks.................................................... 36
4. DATA MANAGEMENT ................................................................................... 38
4.1. Data Acquisition and Mining ............................................................................. 38
3
City.Risks
Deliverable D2.1
4.1.1. Mining the social media ........................................................................................ 38
4.1.2. Sentiment analysis ................................................................................................ 39
4.1.3. Event detection..................................................................................................... 39
4.1.4. Event summarization ............................................................................................ 40
4.1.5. Geocoding ............................................................................................................. 41
4.2. Query Processing .............................................................................................. 42
4.2.1. Identifying and ranking locations and areas of interest ....................................... 42
4.2.2. Queries with spatial, temporal and textual filtering ............................................ 44
4.3. Privacy and Anonymization .............................................................................. 47
4.3.1. Privacy issues and anonymization techniques ..................................................... 47
4.4. Data Analytics ................................................................................................... 49
4.4.1. Mapping crime ...................................................................................................... 50
4.4.2. Predicting crime .................................................................................................... 51
4.4.3. Crime analytics using the social media ................................................................. 51
4.4.4. Crime analysis software ........................................................................................ 52
4.5. Route Planning .................................................................................................. 53
4.5.1. Shortest path queries in road networks ............................................................... 53
4.5.2. Generalised path queries...................................................................................... 55
4.5.3. Generalised routing problems .............................................................................. 57
5. MOBILE SENSORS AND COMMUNICATIONS ........................................................ 60
5.1. Sensors and Communication ............................................................................ 60
5.1.1. Communication technologies for wireless sensor networks ................................ 60
5.1.2. Network topologies and operational modes ........................................................ 61
5.1.3. Mobile wireless sensor networks ......................................................................... 62
5.1.4. Coverage issues for MWSNs ................................................................................. 63
5.1.5. Data management issues for MWSNs .................................................................. 64
5.2. Ground Monitoring and Theft Detection .......................................................... 65
5.2.1. Location finding .................................................................................................... 65
5.2.2. Existing monitor systems ...................................................................................... 65
5.2.3. Collaborative target detection with decision fusion ............................................ 66
5.2.4. Public safety and surveillance networks in cities ................................................. 67
4
City.Risks
Deliverable D2.1
5.3. Mobile Sensing in City.Risks .............................................................................. 68
5.3.1. BLE relevant technologies..................................................................................... 68
5.3.2. Bluetooth low energy technology overview ......................................................... 69
5.3.3. BLE fundamental characteristics .......................................................................... 70
5.3.4. Beacons ................................................................................................................. 72
5.3.5. Theft detection sensor technical consideration ................................................... 74
5.4. Existing Solutions and integrated projects related to City.Risks BLE theft
detection sensor ...................................................................................................... 78
5.4.1. BluVision/CycleLeash concept .............................................................................. 78
5.4.2. Bike Track concept ................................................................................................ 80
5.4.3. BikeTrack fundamental principles ........................................................................ 80
5.4.4. BikeTrack platform components .......................................................................... 81
5.4.5. Customized Bluetooth tag installed on bicycle..................................................... 81
5.4.6. Implementation of client app on mobile phone................................................... 82
5.4.7. Back-end software ................................................................................................ 82
6. SOFTWARE DEVELOPMENT METHODOLOGIES AND PLATFORM ARCHITECTURES .......... 84
6.1. Software Development Methodologies............................................................ 84
6.1.1. Design process models and theories .................................................................... 84
6.2. Platform Architectures for Emergency Management Systems ........................ 88
6.3. Research directions ........................................................................................... 91
7. CONCLUSIONS............................................................................................. 92
APPENDIX I. ROLLED OUT APPLICATIONS ............................................................... 93
APPENDIX II. ONGOING PROJECTS’ DEVELOPMENT .................................................. 99
BIBLIOGRAPHY ............................................................................................. 102
5
City.Risks
Deliverable D2.1
1. Introduction
The City.Risks project aims at increasing the perception of security of citizens in
urban environments. This will be achieved by placing information sharing at the
center of addressing security challenges in large urban environments. Citizens will
serve both as targets and sources of information. Information flow will be
bidirectional between citizens and authorities or among citizens themselves, forming
trusted networks and communities. Smart phones and mobile devices will be utilized
as the enabling technologies for the visualization and acquisition of information. The
City.Risks platform will analyze and integrate diverse information including historical
crime data, statistics, victimization reports, demographic data, maps of
transportation networks, physical sensory data, news feeds and information mined
from the Web.
The City.Risks project will have to address multiple challenges from different
research areas. To do so effectively, we must exploit current research advances and
extend the state-of-the-art to effectively complete the project tasks. The first step
towards this direction is provided by this document, which overviews the most
important work that is relevant to City.Risks. The analysis is conducted both at a
project level and at the level of individual areas of research that are relevant to the
City.Risks project.
The first part of this analysis focuses on the Urban Security Landscape and provides a
structured analysis of complete projects and integrated solutions. In particular, a
total of 22 projects and 46 software and hardware solutions are investigated. Most
of these solutions are the result of relevant research projects. These are analyzed
across a number of relevant dimensions. Depending on the use cases they serve,
these solutions focus on different fields of crisis management. Depending on their
target audience, they may cater for general public, governmental bodies, control
center staff and public safety forces. Depending on their role, these systems may be
designed to provide information, enable communication or collaborations. The study
follows additional distinctions such as the platforms, data sources and the features
offered by the analyzed solutions.
The second part of our analysis presents the state-of-the-art in the areas of research
that are relevant to the City.Risks platform. For every identified relevant area, we
outline the most important recent advances in relation to the tasks that must be
completed during the duration of City.Risks. Furthermore, we present research
directions that will be exploited based on gaps and challenges in the current work.
Emergency Response and Risks Management is a research area directly related to
the project. Of particular significance are the topics of Emergency Alerting and
Response, Mobile Augmented Reality for emergency response and Risk
Management. To that end, we analyze the technology-driven, psychological,
sociological and organizational aspects of alerting. Emergency response is broken
down and analyzed in the levels of response creation, monitoring and organization.
With respect to Mobile Augmented Reality, we focus on user interaction with mobile
augmented reality data, content models, and the generation, provision and
6
City.Risks
Deliverable D2.1
rendering of mobile augmented reality data. Finally, we overview important work
associated with Risks Management, the Operation Centre, and the Simulation and
Visualization of crime information and risks.
Another important area of research to the project is Data Management. Gathering
information from various sources is vital to City.Risks, and to this end we outline the
most important relevant work in Data Acquisition and Mining. We focus on mining
information from the social media, performing sentiment analysis, detecting and
summarizing events, and geocoding web documents. Query processing is also a very
important topic with respect to City.Risks. We present the state-of-the-art in query
processing, and particularly focus on query processing with data with spatial,
temporal and textual characteristics, which are related to the types of data that must
be efficiently managed by the City.Risks platform. Moreover, as some of the data
City.Risks may deal with include potentially private information, we study the
aspects of privacy and anonymization. In addition, we discuss data analytics methods
for mapping and predicting crime, crime analytics using information from the social
media, and we overview the state-of-the-art software used for crime analysis.
Finally, we present work in route planning queries, which constitutes another key
area of focus for the services we plan to develop in the course of the project.
The third research area of focus relates to the theft detection requirements of the
City.Risks platform. This is the area of Mobile Sensors and Sensor Communication.
We overview recent sensor communication advances and we discuss important
aspects for Mobile Wireless Sensor Networks, such as network topologies,
operational modes, network coverage and data management. Particular attention is
given to the topics of ground monitoring and theft. We focus on the problems of
location finding, collaborative target detection and public safety surveillance
networks. Our analysis continues with the mobile sensing technologies that will be
employed in the City.Risks project. In addition, we discuss important technical
considerations for theft detection sensors, and we analyze relevant solutions related
to the sensors that will be employed in City.Risks.
The final research area of focus is Software Engineering. Our analysis reviews current
Software Development Methodologies and Platform Architectures that have been
employed for the development of Emergency Management Systems.
This document is structured as follows. Section 2 overviews the Urban Security
Landscape. Section 3 presents current work in Emergency Response and Risks
Management. Recent advances in Data Management are provided in Section 4.
Section 5 focuses on Mobile Sensors and Communication. Relevant Software
Development considerations are discussed in Section 6. Finally, Section 7
summarizes and concludes the report.
7
City.Risks
Deliverable D2.1
2. Urban Security Landscape
2.1. Gaps and Challenges
For initial research in the City.Risks project, a total of 22 projects as well as 46
software and hardware solutions, partly results from aforementioned projects, were
analyzed and structured in a catalogue for further investigation. This survey of
existing works in this area emphasized on use cases, target audience, solution type,
platform, data sources and addressed features of the solutions analyzed.
Due to lack of available demonstrators provided by the developers, no qualified
statement could be given concerning usability and conformity with standards of the
examined hardware and software.
It is also noted, that many products sold are not more than mere video cameras
streamed to a command and control center of some kind. These were not included
into the deeper analysis, as they would not have created any added value for the
results.
2.1.1. Use cases
As expected, the majority of solutions are set out to tackle public safety and security
threats. Nevertheless, different fields of crisis management are addressed by the
solutions analyzed. These different emphases will be displayed in more detail.
A broad number of the use cases presented is set out to inform citizens or collect
data from their end user devices, employing sensors in the devices or social media
used. Descriptions of solutions, which focus on data collection and information
distribution, state, that during the development a central focus of the research and
development activities was oriented towards ethical, privacy and legal aspects.
Nevertheless, no extensive statements on legal aspects or standardization activities
taken into account are provided by the solution developers.
Other aspects of the development tasks for the solutions analyzed were, on the one
hand, to guarantee mobility of the end users by directing and coordinating them
through centralized communication headquarters, and, on the other hand, to
support policy makers and their strategic decision making.
2.1.2. Target audience
The main target audience of the solutions can be divided into four major categories:
the general public, governmental bodies, control center staff and public safety
forces. The functionalities provided to them are described subsequently.
The general public is primarily addressed by information services and sometimes
included into collaborative actions. Most solutions offer a communication channel to
8
City.Risks
Deliverable D2.1
deliver actions to be taken in accordance with the current emergency situation.
When citizens have the chance to participate i.e. by reporting as a first responder,
they are provided with established social media contact points (most commonly a
Twitter Channel).
Governmental bodies and policy makers are normally provided with analysis and
steering methods. An outstanding feature of the ISAR+ project prototype
(http://isar.i112.eu/) is providing users with well-documented strategic planning
tools.
Communication on various channels and real time information retrieval are the key
selling proposition for control center solutions that were analyzed throughout the
initial research. Common data sources for these solutions are social media or video
data provided by first responders on site.
Public security and safety, especially police forces, are commonly addressed by ways
to monitor suspicious behavior or to track down persons and objects. These
solutions are most commonly based on video surveillance.
2.1.3. Solution type
The majority of solutions analyzed are proprietary systems, requiring special
hardware to be installed (i.e. control center equipment). Those systems, which are
set out to integrate citizen actions into the risk assessment and reduction, are built
to be used on mobile devices, especially cell phones. The mobile solutions can be
subdivided into three main solution types. First, using social media channels for
interaction with citizens is suggested. Secondly, collection of mobile phone sensor
data is employed. Thirdly, information is provided through proprietary web portals
that can be accessed by mobile browsers.
2.1.4. Platforms
As stated before, most systems are proprietary solutions, requiring special hardware
and platforms. Those products and services built for common electronic devices like
desktop computers, tablets and smart phones are equally spread among operating
systems. For desktop computers, this is primarily enabled by using web technologies,
providing online solution to the end users, enabling independence from Windows,
Mac OS or Linux based operating systems. If mobile applications are developed, they
are at least built for Android and iOS. Only few solutions provide further apps for
Windows Mobile and Blackberry operating systems.
2.1.5. Data sources
Public security and safety solutions show three primary sources of input data, being
generated by end device users, being collected by the devices automatically or being
based on statistical data. User generated data is most commonly collected using
9
City.Risks
Deliverable D2.1
social media data aggregation, but also through user reporting tools and knowledge
wikis or encyclopedias, which allow citizens participation and expansion.
Automatically collected data includes mobility, traffic, text, audio and video data,
either collected by the mobile devices of the end users or by sensors developed for
specific purposes.
Besides the aforementioned solutions, which provide real time data, some solutions
are built on previous results only. The sources included into these products and
services are statistical data, maps, infrastructural information and counter measures
definitions.
When analyzing data sources and usage, also the option of reuse of the data was
investigated. Nevertheless, most solutions do not provide any of their data sets. Only
few implementers offer partially access to data or knowledge generated. Those
offers consist either of wikis that were setup during projects and deliverable
documents, which were made publicly accessible, or an API to open data, that was
used for the solution developed.
2.1.6. Features provided
Building on the data collected, the safety and security solutions analyzed provide
diverse features to the end users. Those products and services built on automated
data collection most commonly serve as surveillance and tracking tools, but also
sometimes enable coordination or policy actions through data filtering, visualization
and statistics.
Other solutions, which build on research result and/or end user generated data, are
set out to observe crisis events, to communicate with persons affected and to steer
public safety and security forces. They are built around a command and control
center solution that connects to mobile clients, enabling the required
communication in unidirectional or bidirectional ways. Unidirectional services are
either gathering information from public reports and user data for deeper insight or
communicate information to the persons in the field for coordination. Bidirectional
on the other hand enable collaboration between the command and control center
staff and the essential end users at the crisis hotspot, allowing interactive resolution
through all affected players.
2.1.7. Dedication to standards
Even though a broad spectrum of solutions is analyzed throughout this initial
research, there are hardly any standards addressed or incorporated in the
descriptions of the products and services provided. Nevertheless, some research
projects like ISAR+ (http://isar.i112.eu/) and INDECT (http://www.indect-project.eu/)
report on implementation of their methods with respect to the European
Convention on Human Rights, the European Union Charter of Fundamental
Freedoms, the UN Convention on the Rights of the Child or the Finnish Personal Data
Act.
10
City.Risks
Deliverable D2.1
Considering the fields of application of the solutions analyzed, some technical
standards, like mobile communication, could be expected to be met by the systems,
but in general they are not explicitly stated by the product and service providers.
Only the two UK based solutions Crime Data Repository and Crime Map dedicate
themselves to usage of data based on JSON standard.
2.1.8. Gaps and challenges identified
When considering the results of this initial research, some gaps in the development
could be identified, which are closely related to the challenges to be faced during
future development of the City.Risks environment. Collection of social and end user
generated data should on one hand side be available to others generating leverage
effects. On the other hand side, collection of personal data has to respect data
security laws and privacy rights, with respect to the legal situation in the country of
application.
Furthermore, a strong dedication to standards, especially communication and
encryption should be considered in further research and development, preventing
potential abuse and misuse of data collected and processed by the City.Risks
solutions. This might contradict with the requirements for availability and desirable
platform independency of the software to be developed.
Thus, one of the challenges of further research and development will be to find a
balance between the aforementioned interest conflicts. Establishing the desired
harmony between them will be part of further steps throughout the City.Risks
project.
2.2. Solutions Overview
A research has been conducted in order to find related practices which will shed
some light on the previous and ongoing work in the field. In the following sections an
overview of the current applications found is projected in various fields. Fields
include status of the project, funded by any European program, business analysis
and technical specifications. The following content is aggregated from the URLs
attached to each solution found.
2.2.1. Research fields
For every project highlighted, basic information provided covers the name, a web
link, the funding reference, the type of the developed solution, the platform it is
built for and the role of the solution. Role describes the type of communication that
is enabled through the solution. The Information role therefore represents
unidirectional communication from organization to public, while Communication
represents the inverse direction, from public to organization. As a bidirectional form
of communication, Collaboration enables cooperation of public and organization
based on reaction of the respective partner.
11
City.Risks
Deliverable D2.1
The developed solutions are then presented with a screenshot and a short
description, followed by details on target audience, use cases, features and countries
the solution was piloted in. In a further step, possible transferrable content is
analysed with regard to the City.Risks needs. This covers mapping to the City.Risks
use cases and description of data used as well as data reusability and currently
available access to the application. User and terms of use explains the user access to
the solution, differentiating between active users, using the application as a tool to
gather data and communicate, and passive users, who are restricted to viewing
(limited) content. Furthermore, it is indicated, whether licences are granted by the
provider.
In a final section of the analysis, further information is provided. This covers the
international scale and scalability of the solution, the availability of support
documents, like guidebooks or manuals, and indications towards international
standards that are respected by the solution.
2.2.2. Rolled out applications
Furthermore, the following finished projects, products and services that are built not
only for mobile usage, but also to work on other hardware systems, were explored
during the initial research phase. For more detailed information refer to Appendix I.
NGHOOD
CITIZENS
CROWD
THEFT
Collaboration
Communication
Name
Information
Table 2.1: Rolled out solutions mapped into communication directions and City.Risks use cases
Urban Securipedia
PEP
COMPOSITE
PACT
Alert4All
SUBITO
INDECT
SECUR-ED
THALES Integrated and scalable
12
City.Risks
Deliverable D2.1
urban security solutions
SAMSUNG Urban Security Systems
SAAB SAFE Emergency Response
Selex ES Urban Security Video
Surveillance
Selex ES CITIESvisor
TAS-AGT Urban Security
LL Tech International Urban
Security
ISS SecurOS
ARMOR
Emexis Fuel Tracker
CrimeReports
UKCrimeStats
Crime Map Vienna - Kriminalität in
Wien
Durham Crime map
iSAR+
Opti-Alert
VirtualGuard
BluCop HappstoR
iHound
GadgetTrak
Prey
Comodo Anti Theft
SafeCity
Legend:
THEFT: Theft of personal items
CROWD: Criminal activity in crowded areas
CITIZENS: Assisting and engaging citizens
NGHOOD: Neighbourhood safety
The solutions found are analysed through the communication direction (grey) and
the pre-elementary use cases (blue) defined so far from the City.Risks project. The
description of Information, Communication and Collaboration fields is given in the
13
City.Risks
Deliverable D2.1
previous section (Section 2.2.1). In addition, we map the analysed solutions with
respect to the use cases that we plan to investigate within the City.Risks project. The
complete specification of these use cases will the result of Task 2.4 and will be
reported in the Deliverable D2.4. This mapping provides a deeper insight on the
feasibility of future adaption of elements of the proposed solution into City.Risks
components.
2.2.3. Ongoing projects’ development
Projects, that were of interest during initial research, but not finished before the
delivery date of the current document, are listed below. Their results will be
monitored throughout the further research and development processes of the
City.Risks project. For detailed information refer to Appendix II.
Table 2.2: Ongoing projects in the field of urban security
Name
URL
Short description
Target Audience
Roles
Use case(s)
ATHENA
http://www.projectathena.eu/
Target Audience
first responders, citizens
Role
Collaboration
Use cases
Crisis management, Social media in crisis
management
eVACUATE
http://www.evacuate.eu
Target Audience
general public
Role
Information, Communication
Use cases
Safety/Security
TACTICS
http://fp7-tactics.eu/
Target Audience:
Threat Manager (TM), Threat Decomposition
Manager (TDM) and Capabilities Manager (CM)
Role
Information
Use cases
Legal, Safety/Security
14
City.Risks
Deliverable D2.1
HARMONISE
http://harmonise.eu/
Target Audience:
Mechanisms/Tools for Delivery of Improved Urban
Security and Resilience
Role
Information
Use cases
Safety/Security
Urban Security eGuide
(Inspirational Plattform)
http://www.besecureproject.eu/dynamics//modules/
SFIL0100/view.php?fil_Id=56
Target Audience:
Education, community, general public
Role
Information
Use cases
Safety/Security
Policy platform
Target Audience:
Policy makers
Role
Collaboration
Use cases
Safety/Security, Policy
Urban security Early warning
system
Target Audience:
Policy makers
Role
Information
Use cases
iRISK Urban Vulnerability
Measure
http://create.usc.edu/sites/defa
ult/files/projects/sow/1045/kur
bancreateyear8annualreportkur
banhudoc.pdf
Target Audience:
NC state policy makers, Howard University
students
Role
Information
Use cases
Economic, Social, Safety/Security
RAW Risk Assessment
Workbench
http://create.usc.edu/sites/defa
ult/files/projects/sow/850/hall2
005riskanalysisworkbenchpart1.pdf
Target Audience:
classified, “official use only” or public environment
Role
Information
Use cases
15
City.Risks
Deliverable D2.1
Safety/Security
DPS Deploy
http://create.usc.edu/researche
r/michaelorosz/projects/dpsdeploy-uscdepartment-public-safety-riskassessment-and
Target Audience:
USC Department of Public Safety
Role
Information
Use cases
Safety/Security
MobEyes
Target Audience:
http://lia.deis.unibo.it/Research urban monitoring
/Mobeyes/
Role
Information
Use cases
Safety/Security, Mobility
2.2.4. Solutions in focus
1. iSAR+
Online and Mobile Communications for Crisis Response and Search and Rescue
URL
http://isar.i112.eu/
Project reference
FP7, Agreement no 312850
Solution type
Document | Web | Mobile | Other
Platform
Compatible with all mobile OS
Role
Information | Communication | Collaboration
Application snapshots
16
City.Risks
Deliverable D2.1
Short Description
iSAR+ is a research and development project aiming to definene the guidelines that, in crisis
situations, enable citizens using new online and mobile technologies to actively participate
in response efforts, through the provision, dissemination, sharing and retrieval of
information for the critical intervention of Public Protection and Disaster Relief (PPDR)
organisations, in search and rescue, law enforcement and medical assistance operations.
From 2013 to 2015, the FP7 iSAR+ project addresses Theme SEC-2012.6.1-3: Use of new
communication/social media in crisis situations.
Results
iSAR+ Guidelines - recommendations for citizens and PPDRs for an effective and
efficient use of social media and mobile technology in crisis situations. iSAR+
Technological Platform - platform integrating ICT tools, including mobile and social
media applications, that provide added-value services for citizens and PPDRs in crisis
situations. Based on the citizens’ and PPDRs’ recommendations, the platform is a
validation tool for the iSAR+ Guidelines.
Application technical details
Business analysis
Target audience
Public Protection and Disaster Relief, search and rescue,
law enforcement, medical assistance, general public
Use case(s)
1) Unattended baggage, 2) Metro Station Fire, 3) Toxic
Gas Accident, 4) Plane Crash, 5) Thunderstorm
Feature(s)
Social media and mobile data aggregation and
presentation, information channels.
Pilot(s)
France, Finland, Ireland, Norway, Poland, Portugal,
United Kingdom
City.Risks Use Cases
Theft of personal items | Criminal activity in crowded
areas | Assisting and engaging citizens |.
Neighbourhood safety
Data usage
Source(s) of data
Social Media and Mobile Data
Data reusable
Yes | No
Access to application
End product published
Yes | No
17
City.Risks
Deliverable D2.1
User and terms of use
User entrance
Registered | Open access | Free | Commercial
User passive
Registered | Open
User active
Registered | Open
License
None
Additional information
Scale
International
Supporting documents
Yes | No
Deliverables
Indication to standards
Yes | No | Not clear
Finnish Personal Data Act
2. Opti-Alert
Enhancing the efficiency of alerting systems through personalized, culturally
sensitive multi-channel communication
URL
http://www.opti-alert.eu/
Project reference
FP 7, Agreement no. 261699
Solution type
Platform
Compatible with all mobile OS
Role
Short Description
18
City.Risks
Deliverable D2.1
The Opti-Alert project is aimed at raising the efficiency of alerting systems through
personalized, culturally sensitive multi-channel communication. Funded by the European
Commission, Opti-Alert involves research institutes, universities, enterprises, and end-users
from six European countries. The goal of the project is to create an adaptive alerting system
that allows intuitive, ad-hoc adaptation of alerting strategies to specific alerting contexts.
Opti-Alert will also facilitate improved regionalization and personalization of warning
messages, as well as a closer cooperation and integration of industry-funded alerting
systems with state-funded alerting tools. The project objectives will be pursued by the
following key research activities:
In-depth analysis of the effect of social, cultural and regional factors on risk
perception and risk communication
Analysis of the influence of the observed socio-cultural differences on regional
alerting strategies
Analysis of the impact of individualized alerting (by SMS, e-mail etc.) and alerting
through broadcast media
Identification of best practices in alerting through broadcast media
Definition of algorithms to simulate alert propagation in the population, both at
large and inside critical infrastructures such as metro stations, as a function of
interpersonal communication patterns and the selected mix of alerting channels.
Results
The main results of this projects are: an Alerting Strategy Simulation, Channel and Broadcast
Suggestion, and Media Broadcasting
Business analysis
Target audience
General public
Use case(s)
Safety and Security
Feature(s)
Alerting Strategy Simulation, Channel and Broadcast
Suggestion, Media Broadcasting
Pilot(s)
Austria (Eastern Tyrol), Germany (Berlin), Italy (Eastern
Sicily), Germany (Lippe)
areas | Assisting and engaging citizens |
Data usage
Source(s) of data
Data reusable
Current emergencies from national authorities
responsible for alerting, weather alerts, disaster alerts,
commercial alert through service providers
Yes | No
19
City.Risks
Deliverable D2.1
Yes | No
User entrance
User passive
Registered | Open
User active
Registered | Open
License
Not known
Scale
International
Yes | No
Deliverables
3. iHound
iHound Phone & Family Tracking Platform
URL
https://www.ihoundsoftware.com/
Project reference
No reference
Solution type
Platform
IOs, Android
Role
20
City.Risks
Deliverable D2.1
Short Description
iHound uses the GPS and WiFi, 3G, or Edge signals built into your devices to determine its
location. Using the app and iHound Software's unique tracking website, you can Track the
location of your device, Remotely Lock Your Phone, Remotely Wipe Private Information,
Directly Instant Message Your Phone, Set up Geofencing Location Alerts, Manage your
account using iHound's Mobile Web Site.
With iHound Geofencing automatically one can:
Get alerts when your child arrives at or leaves school.
Broadcast your arrival at the bar to your adoring Facebook and Twitter followers.
Check-in with Foursquare to help you compete more effectively and become Mayor.
Receive reviews of local restaurants when you arrive at your vacation destination.
Have your shopping list delivered in an email when you get to the store.
Know where your loved ones are all the time, any time.
Business analysis
Target audience
General public
Use case(s)
Safety, Security, and Mobility
Feature(s)
Geofencing, Location Tracking, Location Sharing, Push
Alert with Sirens
Pilot(s)
Not known
21
City.Risks
Deliverable D2.1
Data usage
Source(s) of data
User data, location based information
Data reusable
Yes | No
Yes | No
User entrance
User passive
Registered | Open
User active
Registered | Open
License
Commercial licence
Scale
International
Yes | No
4. GadgetTrak
Leading innovator of theft recovery and data protection solutions for mobile
devices
URL
http://www.gadgettrak.com/
Project reference
No reference
Solution type
Platform
iOS, Android, Win
Role
22
City.Risks
Deliverable D2.1
Short Description
GadgetTrak Laptop Security
Think about the things you have on your laptop right now: Countless photos, financial
records, software, music, videos, etc. The hefty price tag on your laptop is probably dwarfed
by the value of the information on it. The really scary part is that according to the FBI, 1 in 10
laptops purchased today will be stolen within the next 12 months. Sadly, only 3% will be
returned. GadgetTrak dramatically increases the likelihood of finding your laptop, by
pinpointing its location, and even sending a photo of thief.
GadgetTrak Mobile & iOS Security
What if your device was lost or stolen, what would it mean to your or your business? Not
only is it the loss of an expensive device, but also your data, which can be priceless.
GadgetTrak Mobile Security helps mitigate the risk of mobile device loss or theft,
empowering you to track its location, back up data, even wipe the device.
Results
GadgetTrak is offering 3 types of security products laptop security, Mobile Security and iOS.
Business analysis
Target audience
General
Use case(s)
Laptop security, Mobile security for data loss and
antitheft
Feature(s)
Tracking, Memory Erasure, Wi-Fi positioning, Integrate
police reports, webcam support, privacy safe, Advanced
hybrid positioning, device alarm, Secure encrypted
backup, Remote data wipe, Tamper proof
23
City.Risks
Deliverable D2.1
Pilot(s)
Not known
Data usage
Source(s) of data
Data reusable
Yes | No
Yes | No
User entrance
User passive
Registered | Open
User active
Registered | Open
License
Commercial license
Scale
International
Yes | No
Wiki, FAQ and Knowledge Base
5. SafeCity
Handy tool to geographically mark and report a safety issue faced by women and
children
URL
Project reference
https://play.google.com/store/apps/details?id=com.
phonethics.safecity
No reference
Solution type
Platform
iOS, Android
Role
24
City.Risks
Deliverable D2.1
Short Description
Safe City is a handy tool to geographically mark and report a safety issue faced by
women and children. You can quickly locate, identify and pin the issue on the map of
your city. It is aimed at creating a social response system where this information is
actively pursued in reaching out to the person in need and bring to the notice the
issues and their causes so they can be addressed and monitored.
Business analysis
Target audience
General public
Use case(s)
Safety, Security, and Mobility
Feature(s)
Location, Danger Spotting, Search
Pilot(s)
Not Known
25
City.Risks
Deliverable D2.1
Data usage
Source(s) of data
Data reusable
Yes | No
Yes | No
User entrance
User passive
Registered | Open
User active
Registered | Open
License
None
Scale
International
Yes | No
26
City.Risks
Deliverable D2.1
3. Emergency Response and Risks Management
This section summarizes the most important work from the literature that is related
to emergency response and risk management services and applications envisaged for
the City.Risks platform. We describe recent advances in emergency alerting and
response, augmented reality, and operation center and risks management.
Furthermore, we pinpoint gaps and challenges in current work to outline specific
research directions that will be explored within the scope of the City.Risks project.
3.1. Emergency Alerting and Response
City.Risks uses different means of information sharing to address security challenges
in large urban environments. The main goals of the project are to improve handling
of safety and security challenges by integrating the public through mechanisms of
participatory urbanism, and making safety-related information more transparent in
order to reduce fear of crime of citizens.
Targeted emergency alerting and response play a major role in this context. Within
this scope, the following sections focus on communication between authorities and
the public. Alerting and response communication used by professional has not been
considered here.
3.1.1. Alerting
After the end of the "cold war", investments in alerting systems for the general
public were reduced in many European countries. As a consequence, the previously
existing alerting infrastructure (sirens in particular) was either dismantled, or its
coverage and / or availability decreased due to lack of maintenance. In the
meantime, authorities mainly relied on mass media (TV, radio) to close the gap on
the last mile when alerting the public. More recently, however, attempts have been
made to use other technologies to communicate with citizens in emergency
situations. This relates both to single-channel approaches (like cell broadcasting
(Jagtman et al. 2010), which is currently the preferred solution in the Netherlands),
or multi-channel alerting systems which combine, for example, SMS, e-mail, and RSS
feeds (Klafft et al. 2008).
Efficient alerting encompasses aspects of the following categories.
-
technology-related aspects
psychological and sociological aspects
organizational aspects
Many of these aspects are interrelated; technology influences how we perceive
alerts, and organizational aspects influence whether we trust emergency
information received.
27
City.Risks
Deliverable D2.1
Technology-driven aspects. According to Klafft (Klafft 2013) a lot of research has
been conducted to analyze the first two steps of the alerting process, namely,
sending and receiving messages. The following technologies are being used.
-
cell broadcasting
SMS and pagers
push infrastructures for mobile phones
twitter
rss feeds, email, fax
others, e.g., tv or radio broadcasts, sirens, sound trucks (not considered)
Cell broadcasting. All available mobile communication standards (GSM, UMTS, LTE)
define broadcast or multicast services. Broadcast services have the advantage of
reaching an affected public without the need of a subscription service. On the other
hand, messages are not explicitly called for and cannot be personalized, which
reduces acceptance. While GSM cell broadcast is actually available and being used
for distributing emergency information in some countries (e.g., US, Japan, or the
Netherlands) usage of these broadcast services is limited, due to the lack of the
necessary infrastructure. Beyond emergency alerting there are currently no real
business cases, and mobile communication providers shy away from investing into
these technologies.
SMS and pagers. SMS and pager technology have been used as a means for efficient
personal alerting, since the advent of mobile communication networks. Pager
communication requires specific devices and separate infrastructures, which limits
its usefulness for alerting the public. In contrast, SMS can be used by (almost) any
mobile phone, and is one of the most widespread technologies used for public alerts.
However, both technologies offer limited information capabilities as they support
only short messages, which is, why they are being more and more replaced by IPbased push infrastructures for smartphones.
Push infrastructures. One of the main technologies currently used for alerting are
smartphone Apps providing push notifications. All major smartphone platforms
provide a push service infrastructure (e.g., Apple Push Notification Service (APN),
Google Cloud Messaging (GCM), Windows Notification Service (WNS)). Push
notifications have the advantage over SMS that they are bound to a smartphone App
that can directly load additional information related to an alert, thus giving a user
details on what happened where, and on how to react. Another advantage of
alerting Apps is that alerts can be subscribed to in a very dynamical manner, using,
for instance, location information.
Twitter. In 2013 Twitter launched the service Twitter Alerts. Tweets that are marked
as an alert are being actively pushed to all users subscribed to alerts of the issuer of
a tweet. That means, subscriptions are based entirely on the who is providing an
alert. Twitter-based alerts cannot be personalized or targeted to specific needs or
user groups. The service is being used, for instance, by the Federal Emergency
Management Agency FEMA in the US (fema.org) or the WHO Regional Office for
Europe (euro.who.int).
28
City.Risks
Deliverable D2.1
RSS feeds, email, fax. All of these technologies have more or less been replaced by
other means. Today, RSS feeds are no longer being used for actual alerting but for
reporting emergency cases, e.g., for earth quake information. Also today, email and
fax are playing only minor roles in alerting the public.
Others. TV or especially radio broadcasts have long been the only means to
distribute detailed emergency information. Also these technologies lack the ability to
target and personalize alerts. New digital broadcast standards (e.g. DAB+), however,
provide means to distribute alerts similar to GCM cell broadcasts. Devices to receive
and display such alerts are available but not actually widely used. In recent years, socalled smart TVs are being more and more available, which means that Apps and
push notifications can be used with this medium, too.
Psychological and sociological aspects. Psychological and sociological aspects (as
addressed, e.g., by EU project OptiAlert (http://www.opti-alert.eu) are a more or less
disregarded topic in existing alerting systems. These aspects address the following
questions.
-
was an alert read
was an alert understood
was an alert reacted upon (see Sec. 3.1.2)
None of these questions can be easily answered by just using the right technology.
Even actually displaying alert information on a device does not mean a user has read
it much less understood its impact. Often, understanding problems arise from the
information being delivered in a single language. Since responsible parties most
often require an alerting system to be generic, alerts are not always built from predefined pre-translated building blocks but using free text input. Automatic
translation can help that information is roughly understandable, but is not good
enough to ensure that the information is being trusted. Trust, however, is the main
point in bringing a recipient to react upon an alert. Trust builds itself on different
components. The source itself needs to be trusted. It must be clear to the user that
the message actually comes from that source and has not been tampered with. The
contents of the message need to be verifiable, preferably using secondary sources
(e.g., social media, online news, radio or TV broadcasts). Last, not least, users need
to have the sense, that they are actually affected and need to react. Individualized
and situation-related alerting is of major importance.
Organizational aspects. Emergency alerting for the public, usually, encompass
information about different types of hazards (e.g., weather, floods, civil protection,
crime, or defense). In most cases responsibilities for alerting the public about such
hazards are distributed between different authorities.
On the other side, emergency situations often arise from a series of interconnected
and dependent severe events. Alerting the public in such a situation requires a
consistent voice which in turn requires a distributed incident management. Also
emergency situations do not stop at political boundaries, so coordination of cross
border alerting strategies (see OptiAlert (http://www.opti-alert.eu)) is of major
importance.
29
City.Risks
Deliverable D2.1
These problems are much more pronounced in federal environments, e.g., in
Germany (where the responsibility for civil protection is divided between ca. 400
individual counties).
3.1.2. Response
Appropriate response is the ultimate goal of all alerting systems, be it that people
look for a safe place, secure their property, help others, inform others, or just stay
informed themselves about their situation. The following section focusses on
collaboration and organizing response between authorities, professional emergency
relief organizations, and the public.
Three main categories or levels can be used to distinguish approaches to emergency
response with focus on the public.
-
creating response
monitoring response
organizing response
Creating Response. Creating response encompass all measures taken to ensure an
alerted public actually takes action. This overlaps with the issues on psychological
and sociological aspects of alerting presented above.
Monitoring Response. Monitoring tries to observe behaviour of the affected public
and to verify users respond (correctly). In this case sentiment analysis techniques
play a major role. Social media content streams the data items of which are tagged
with temporal, spatial or keyword-based metadata is processed in order to get
information how an emergency situation is perceived and reacted upon.
The role of social media (read twitter) in effectiveness of warning response (with
focus on extreme events) has been studied in (Tyshchuk 2012). Twitter users seem
to engage in all six stages of the warning response process (receiving, understanding,
trusting, personalizing, obtaining confirmation, taking action).
Enhancing situational awareness through the use of microblogging (read twitter) has
also been the focus of (Vieweg et al. 2010) and (Starbird and Palen 2011).
Organizing Response. Organizing response takes additional steps to facilitate
collaboration between authorities, emergency relief organizations, and the public,
through direct or indirect communication.
Effective response and recovery operations through collaborations and trust
between government agencies at all levels and between the public and nonprofit
sectors has been studied by (Kapucu 2005) and (Kapucu 2008). Key aspects of the
relationship between volunteers and formal response organisations in disasters are
presented by (Barsky et al. 2007). (Chen et al. 2008) proposes a framework to
analyze coordination patterns along the emergency response life cycle.
Several systems have been developed to help coordinate responder communication
and response efforts in order to minimize the threat to human life and damage to
property. Examples include the ENSURE (http://ensure-projekt.de/) project and
30
City.Risks
Deliverable D2.1
(Yuan and Detlor 2005). Another flavour of systems around cooperatively responding
to "threats" is community-based neighborhood watch systems.
One measure that sticks out of this category is running disaster exercises. Disaster
exercises can also be seen as a means to create the right response in an emergency
situation, it, however, needs a complete other level of organization and direct
collaboration between authorities and citizens then the measures addressed above.
The role of such exercises to improve effectiveness in emergency management and
in creation of community disaster preparedness, has been laid out by (Perry 2004).
Exercises result in increased knowledge of the participants (what to do in an
emergency) as well as increased confidence in abilities of others and in the ability to
work as a team.
3.1.3. Research directions
In contrast to broadcasting, targeted alerting requires information about the
situation of the group or individuals to target. One of the challenges is to preserve
the privacy of the users. There are different techniques that can be used to achieve
this.
Pseudonymization of data records, using random device identifications
instead of personal user ids, like, email addresses
Encryption of data records, e.g., using Bloom filters to guard against
unauthorized access while maintaining searchability of data sets
Decentralized storing of data records, e.g., using personal or on device space
to hold sensitive data
Since City.Risks tries to target a wide spectrum of use cases, alerting and – in
particular – response functionalities require a lot of flexibility. In most cases, great
flexibility leads to losses in scalability and overall performance of systems. To
prepare City.Risks platform components to be used in real products, these
flexibility/performance trade-offs has to be taken into account, to be able to serve
huge number of users simultaneously as a smart city environment requires.
3.2. Mobile Augmented Reality (mAR)
The City.Risks mobile application offers new visualization methods for
communicating (individual) risks as an effective innovation for better action advices
by integrating and developing mobile augmented reality (mAR) technologies. This
mobile application with augmented reality features will allow citizens to use their
smart phones or tables to visualize security-related information regarding their
surroundings (e.g. historical crime statistics or any ongoing criminal activity), being
also able to interact with the augmented objects (e.g., for tagging them). This will
enable better situation-awareness and response support.
31
City.Risks
Deliverable D2.1
3.2.1. User interaction with mAR data
Mobile augmented reality (mAR) is an innovative user interface for providing
individual risk information and warnings as well as action advices on mobile devices
(e.g. smartphones). Individual risks and defence action alternatives can be presented
more realistic, e.g. by presenting the virtual water level of flood (risk map) or
potential danger defence actions (alternative kinds of level) directly in reality, which
improves the individual recognition and analysis of a hazard situation as well as
possible action possibilities.
The data presented as augmented reality can include information about visible
objects (e.g. the statics of a building), not visible objects (e.g. a below-ground
infrastructure like electricity or sewage system) or not yet visible objects (e.g. a
flooding). For example, users can see information about their surrounding location
regarding history of criminal activity (crime maps by type and seasonality), as well as
any ongoing criminal activity or a coming criminal activity in the near future (e.g. a
group of terrorists moving to the users location).
Today, mAR applications only present information (often 3D objects) to the user. But
in case of an ongoing or future criminal activity (like in City.Risks), the
communication should turn to a bidirectional channel, with the users (witnesses)
sending back feeds to the authorities. The challenge is to provide users the
possibility to interact with the augmented objects, e.g. to add information, to update
objects or to create new objects. Similarly, methods and tools for capturing of new
content by the users (crowd sourcing, in-situ authoring) are not available (research is
focused on authoring tools for expert authors, e.g. DART, arToolkit, Layar Creator)
(Chun et al. 2013) (Rumiński et al. 2013); thus, research will be conducted in the field
of design of interaction with spatial mAR content.
3.2.2. mAR content models
Traditionally, only a single risk is presented as AR in a mobile application. An
additional challenge that will be addressed in the City.Risks project is to visualize a
set of (all) risks (historical as well as current as well as future) in a single mobile app.
Different data sources (data bases, web services, GIS etc.) with different data types
and formats will be integrated and mapped to a common content model (Rumiński
et al. 2014). Unfortunately, no standard AR data format currently exists.
Current mAR content standardization (OGC ARML 2.0 (Lechner 2013), AR Standards
(Perey 2014), AREL (Metaio 2014) and X3D AR (SRC WG 2013)) do not address the
problem of a common content model for a consistent mapping and management of
heterogeneous content from different data sources, because they only focus on a
data format for visual presentation of mAR content. Therefore, research for such a
common data exchange format is necessary and different available data formats
have to be investigated to develop such a common data exchange format.
Another challenge is the integration of spatial data, its combination with additional
data, and the subsequent generation of appropriate 3D structures. Most promising is
the XML-based OGC standard CityGML as such a mAR data exchange format. Hereby,
32
City.Risks
Deliverable D2.1
the spatial data objects (as connecting elements of the real environment as well as
the augmentation) could be mapped with its structure, geometry and semantic in
appropriate structures of the standard CityGML (Gröger et al. 2012) which could be
transformed in an appropriate 3D structure: mapping of the actual geometry (e.g. for
building); representation be (realistic) symbols (e.g. vegetation); approximation by
extrusions. These information can be combined with other spatial data (e.g.
aerophoto for texturing), so that from the existing spatial data sets result
comprehensive descriptions for the mAR Scenes.
The use of the standards CityGML probably enables – besides the mapping of the
geometries – also the preservation of the structure of the objects and the mapping
of the semantic, which can be used as a basis for the subsequent scene authoring.
The building of the 3D structures will take place rule-based, c.f. to the greatest
possible extent automatically.
3.2.3. Generation, provision and rendering of mAR data
The information provided to the citizen about the individual risk at a certain location
can be static (maps, crime statistics) or dynamic/real-time (current events,
conditions). As relevant data (e.g. real-time crime maps, flood risk maps) can be
complex geometries, existing approaches of web services for data generation,
provision and rendering for mAR on mobile devices (Belimpasakis et al. 2010) are
made for static data and are based on caching (e.g. tile map services, TMS), which
are too inflexible in dynamic danger situations (storage use, performance, user
experience UX).
A challenge is to provide dynamic complex mAR data in real-time. The main
approach will be to develop a dynamic, cloud-based, OGC conform Web Processing
Service (WPS) within the existing holgAR content platform (Fuchs-Kittowski et al.
2012), (Fuchs-Kittowski et al. 2014) for providing data necessary, that uses clipping
to reduce the complexity of a polygon (e.g. risk map). By clipping, a specific area is
cut out of a risk map to create a smaller map (a less complex geometry with less data
to process). This way, the generated geometry is small enough to be rendered on
demand dynamically. In addition, outsourcing resource intensive tasks to the cloud
(e.g. Google App Engine) (Huang et al. 2014), (Manweiler 2014) will be used to
ensure a high performance, reliable and robust system in case of a disaster, where a
large number of requests are expected.
During the design of the City.Risks mobile application, we direct our research to new
and vibrant topics related to the use of augmented reality in security solutions.
Development of a mAR data description language (mAR content model): As a basis
for a consistent, neutral representation of spatial content from different data
sources in different formats a mAR content description language will be developed.
This consistent data model also enables the generation of different export formats
33
City.Risks
Deliverable D2.1
ranging from simple formats (e.g. points of interest) to complex objects (2D area, 3D
structures etc.), e.g. AREL for presentation in AR browsers. The technical basis will be
the XML-based standard CityGML, which is well established in the world of spatial
data and will be investigated and probably used in the project. Particularly, CityGML
provides a good basis for the description of geometries and semantic and is – as a
generic language – extensible, e.g. for flood, crime and other risks.
Development of mAR dynamic, cloud-based WPS: Development of a dynamic, highperformance, cloud-based, OGC conform Web Processing Service (WPS) for the
generation and provision of dynamic mAR representations in real-time. It will use
techniques like clipping to render a complex geometry (e.g. a risk map) on demand
dynamically, and will be implemented as a cloud service to ensure a high
performance, reliable and robust system in case of a risk, where a large number of
requests are expected.
Development of an user interaction concept for spatial mAR content: An
interaction concept for spatial mAR content will be developed and implemented to
present actual risk information as mAR in the camera view of the mobile device in
the context of reality and allow the users to interact with it. It will provide users of
the mobile app the possibility to interact with the virtual, augmented objects
presented as AR, e.g. add information, update objects or create new objects.
3.3. Risks Management and Operation Center
In the City.Risks project, the risk management and operation center form the
command and control center. These systems are triggered when citizens transmit
information from a scene or the sensor–based theft detection module reports stolen
objects. Further they are used for monitoring of crime incidents and visualization of
information from “ground reporters”. Further data simulations and evacuation
models are made based on the incoming data. An operation manager will have
access to all the combined components in order to manage and control the
processes.
3.3.1. Risk management
Risk management can be seen as the super ordinate term behind crisis management,
disaster management, emergency management, conflict management, and incident
management. Risk management provides the broadest context of warning systems.
(UN-ISDR, 2009) defines risk management as “managing the uncertainty to minimize
potential harm and loss”. The main goal of risk management is to develop a strategy
as a measure for reducing risk. Important aspects of risk management such as risk
and vulnerability analysis could be found as constituting elements of crisis warning
systems. The use of the term warning system in the context of risk management is to
monitor indicators and alerting within smaller scale systems such as for road safety
(Banks et al. 2009), business logistics, or IT systems thus on risks that cannot be
assigned to more specific management domains.
34
City.Risks
Deliverable D2.1
The risk management system reacts when sensors or citizens transmit new data. The
different information like pictures, videos, notes and theft detection messages are
analyzed. An alarm manager will then react on the data. (Hollifield and Habibi 2010)
developed seven steps to alarm management. Further, risk assessment is done to
identify in advance what could happen and how to react. The Urban Risk
Assessment (URA) is a study done by (Dickson et al. 2012) and analyses disaster and
climate risks in cities. They present a flexible approach to assess the risks. (Shi et al.
2006) looked at the urban risks especially in China.
Recent research in the field of risk, disaster and emergency management has
proposed reference architectures for the open composition of risk management
systems based on Software as a service (SOA) principles and proven concepts (e.g.,
the concepts from the projects ORCHESTRA, OASIS and WIN (Sassen et al. 2005),
SANY (Havlik et al. 2006) or INSPIRE (Florczyk et al. 2009).
Often the operation center is seen as a part of the risk management, but it could also
be seen as separate part working together with the risk management.
3.3.2. Operation center
The operation center and risk management system is a physical or virtual facility site
from which response teams exercise reaction and control in an emergency case,
disaster or security incident. In a city, different providers take care of the security of
the citizens such as the fire brigade, police department or health care services.
Further there exist operation centers for critical infrastructures, like energy suppliers
or transport services. All in common monitoring, coordination and managing are the
main tasks for an operation center. They work for daily threats, natural disasters, in
case of defense or civil protection. Current research is clearly directed to an
operation center that offers situation awareness over the whole city and all related
aspects. This direction is driven by the smart city concept.
Smart City Operation Centers: With more and more people living in a city, new
requirements arise to respond also to crime situations. For example, New York City
reduced crime by 27% with a centralized location of information and a 911 real-time
dashboard providing emergency needs and its resources (Washburn et. al 2009). The
Cisco Smart+Connected City Operations Center also offers Integrated Control, Input
collection, transmission and distribution as well as multi-display operation (Cisco,
2015). The IBM Intelligent Operation Centre is built for cities and communities to
integrate systems and information sources, tools for communication and co-working
and analysis and visualization (IBM, 2012).
(Naphade et al. 2011) describe the smart cities and their future challenges. Public
safety is one key factor besides education, healthcare, transportation and others.
The operation center of Rio is described as an example of an integrated information
space. As seen in Figure 3.1, they include dynamic information from weather
sensors, video surveillance and others and combine them with GIS data to support
the crisis and transportation management. For the future, more services should be
connected to the system in order to have a closed-loop system.
35
City.Risks
Deliverable D2.1
Figure 3.1: Rio de Janeiro’s Operations Center after (Naphade et al. 2011)
3.3.3. Simulation and visualization of crime and risks
A key requirement of the operation center is to quickly get an overview about all
incidents and information. This is supported by simulation and visualization. To
visualize crime and risks, crime mapping is done. On the basis of Geographic
Information Systems (GIS) and temporal aspects, it is possible to localize and analyze
all incidents. Originally used by criminologists to combine all available data to one
case, it could also be used to map specific information from different cases. Hot
spots of crime could be visualized in order to identify areas of risk. Two different
aspects have to be considered for the visualization, the geographic areas looked at
and the color used. (Harries 1999) suggested the usage of point, lines or polygons.
Points represent exact places with addresses or street corners, lines represent
streets and paths, which could be straight, bent or curved and polygons represent
neighborhoods as a bigger area in a city. To visualize the density of incidents (Eck et
al. 2005) presented the usage of graduated dots so that the size represents the
number of crimes. Further they proposed to use color gradient dots to visualize
nuances for example a coloring from red to yellow.
(Boulos et al. 2011) use crowdsourced data from Twitter and Wikipedia to compose
a 3D visualization or outdoor surveillance data in real-time. Sensor data and citizens
reports are combined for the team of professionals to react in the operation center.
In Figure 3.2, the architecture using the data for the visualization is shown.
36
City.Risks
Deliverable D2.1
Figure 3.2: Common Scents technical architecture after (Boulos et al. 2011)
The City.Risks platform will follow the resent research directions and integrate
different components, usually built as autonomous parts to work as one system.
Important parts are the communication between the single functions, data
messaging as well as data formats and protocols have to correspond.
The operation center will allow the operator to have an overview of all incidents and
threats in order to respond to them. A visualization tool will provide all the
information on a map in order to comprehend them quickly and calculate the
situation. Also the usage of social media and the possibility of the communication
between the operator and the citizen is a new field. The participation is essential for
the operator to get details of the situation. The decision making process is enhanced
by the risk management and the simulation platform.
37
City.Risks
Deliverable D2.1
4. Data Management
This section overviews the most important work from the literature that is related to
the Data Management layer of the City.Risks platform. We describe recent advances
in data acquisition and mining, query processing, privacy and anonymization, data
analytics and route planning. Furthermore, we pinpoint gaps and challenges in
current work, and outline specific research directions that will be explored within the
scope of the City.Risks project.
4.1. Data Acquisition and Mining
The City.Risks platform must facilitate decision support and data analytics tasks with
the focus on security and the reduction of the fear of crime. In order to effectively
complete this task, the exploitation of data from multiple sources that describe the
urban setting is necessary. To that end, City.Risks will not only utilize standard
datasets, such as official demographic data or police crime reports, but will also
focus on the integration of data from additional sources, such as the web or the
social media, in order to enhance our understanding of crime related information in
urban environments.
4.1.1. Mining the social media
Social networks have gained significant attention over the past few years. They have
attracted a worldwide user base that exchanges vast amounts of information while
communicating, sharing thoughts and describing real world happenings. Twitter
(Kwak et al., 2010) in particular has been in the forefront of such activity, attracting
also significant research interest. The open access to Twitter data and its real-time
nature makes information sharing instant. These characteristics have directed a large
part of the research in mining the social media to focus primarily on Twitter data.
Most of these approaches can be utilized with respect to similar data from different
sources. Effectively mining Twitter data (Bontcheva et al., 2012) can lead to valuable
information about real-world phenomena, such as real-world emergencies. Recent
works include the usage of Twitter for crisis management (Cameron et al., 2012) and
analysis of information diffusion in the case of natural disasters (Mendoza et al.,
2010).
Mining Twitter data is more challenging compared to extracting meaningful
information from extensive, structured text usually found in news reports. Tweets
have limited size, restricted to 140 characters. Such user-generated content is
usually noisy, and has no imposed structure, often characterized by informal
language and syntax, as well as extensive use of abbreviations, special characters
and emoticons. In addition, its worldwide user base results in multilingual content.
To overcome these issues, multiple methods have been proposed (Bontcheva et al.,
2012). Natural Language Processing (NLP) and Machine Learning (ML) methods have
38
City.Risks
Deliverable D2.1
been employed to analyze the content, ontologies have been developed to
represent semantics, and graph theoretic methods have been used to analyze the
structure of the network formed by the Twitter users and their relations.
4.1.2. Sentiment analysis
Sentiment Analysis and Opinion Mining (Dave et al., 2003) is the process of
identifying and extracting subjective information. Using sentiment analysis, one can
monitor what people have in mind in terms of sentiment (“good”, “bad” or
“neutral”) when they refer to a specific subject. This is commonly achieved by
analyzing text and measuring the sentiment for the given text snippet using NLP
techniques.
One interesting ongoing research direction is the enhancement of sentiment analysis
techniques in order to detect moods and emotions in tweet content. For example,
(Bollen et al., 2011) mines the public sentiment (positive / negative) and mood
(calm, alert, sure, vital, kind, happy) from tweet feeds, and investigates the
effectiveness of such information towards predicting stock market valuations.
Another recent work (Lansdall-Welfare et al., 2012) applies standard mood detection
methods from (Strapparava and Valitutti, 2004) to detect sentiments (joy, fear,
anger, sadness), and correlates public mood patterns in the UK to detect periodic
events such as Christmas and Halloween, and public events such as riots and the
announcement of cuts in public spending.
4.1.3. Event detection
As mentioned above, the real-time nature of Twitter allows the instant propagation
of information about real-world events as they happen. This characteristic allows
Twitter users to be more responsive in diffusing information about real-world
happenings than curated news feeds or other social media.
Event detection from Twitter data has attracted attention recently. One of the most
influential works is that of Takaki et. al (2010), which focuses on the detection of
earthquakes. Twitter users are considered as sensors, and tweets as sensory
information, reducing the event detection problem to object detection and location
estimation. By using machine learning techniques, tweets that are relevant to
earthquake events are identified. Then, the content of tweets, as well as geolocation information is employed to detect and locate events.
Twitter data have been also employed in order to detect crime and disaster related
events. In (Rui et al., 2012), a classifier determines if a tweet is related to such an
event. The approach is based on a set of rules that outline crawling strategies for the
detection of relevant tweets, which are refined iteratively. Apart from the text of a
tweet, other features, such as links in the content, serve as input to a classifier.
Location prediction is based on the fusion of information extracted from the Twitter
graph, as well as the knowledge gathered from past tweets and re-tweets.
39
City.Risks
Deliverable D2.1
Gerber (2014) utilizes twitter data and kernel density estimation in order to predict
crime. The output of twitter-specific linguistic analysis in combination with statistical
topic modeling is used to generate additional inputs for a crime prediction model.
The author presents experimental results indicating that, for 19 out of the 25 types
of crime, the addition of the Twitter data input increased the performance of the
crime predicting method compared to a standard kernel density estimation based
method.
Weng and Lee (2012) construct signals for individual words by applying wavelet
analysis. Trivial words are filtered out according to their corresponding signal autocorrelations. Events are detected by clustering the remaining words using a
modularity-based graph partitioning technique.
Becker et al. (2011) analyze the stream of Twitter messages with the purpose of
detecting which messages refer to real-world events, and do not refer to Twittercentric trending topics. Their approach relies on clustering of tweets that are
topically similar.
Contrary to most works on event detection that focus on large scale public events,
Agarwal et al. (2012) investigate the detection of events of local importance that are
usually sparsely reported. Their approach is based on a two-step process, where
supervised classification is employed to detect relevant tweets, and NLP is applied to
analyze tweet content.
Recent work by Nathan Kallus (2014) investigates the quality of Twitter data towards
the prediction of events. Off-the-shelf NLP tools are used for the processes of event,
entity and time extraction. A case study of this approach is the prediction of protests
in a country and city level. Precision is evaluated against event extraction performed
by workers on Amazon Mechanical Turk.
4.1.4. Event summarization
Apart from the detection of events, other works are focused on the summarization
of known events based on data mined from Twitter. Chakrabarti and Punera use
tweets to summarize sporting events. They formalize the problem of summarizing
events using tweets, and propose a solution based on learning the underlying hidden
state representation of events using Hidden Markov Models.
In order to utilize information from a plethora of sources to better detect and
summarize real-world events, recent works focus on the fusion of data retrieved
from multiple social media. This requires the aggregation of different types of
content, such as text, images and video. (Becker et al., 2012) combines precisionoriented and recall-oriented query generation techniques to associate content from
different social media with events, and as a result utilize content extracted from one
social media site in order to enhance the detection of events in other social media
sites. Their methods have been applied to detect information from Twitter, Flickr
and Youtube.
40
City.Risks
Deliverable D2.1
4.1.5. Geocoding
Spatial location is an important aspect of real-world events, which may be crucial in
the decision support process. To that end, multimedia geotagging is an important for
a variety of applications (Luo et al., 2011). With respect to the City.Risks domain, the
detection of crime events is incomplete in the absence of information about the
spatial location that the criminal activities took place.
Usually web documents are not associated with spatial coordinates. Even with
respect to platforms that allow the geolocation of documents on creation using GPS
coordinates, only a small fraction of the users utilize this feature. For instance, only
0.77 percent of public tweets include GPS coordinates (Semiocast, 2012).
The task of automatic document geolocation involves the resolution of spatial
locations described within textual descriptions. This process is usually achieved by
extracting spatial named entities from a document and assigning spatial coordinates
to these entities using a gazetteer. Domain specific applications utilize additional
heuristics in order to increase the process effectiveness.
NewsStand (Teitler et al., 2008) monitors, collects and displays news stories in a map
interface. A custom geotagger is used in order to obtain location coordinates
associated with the articles, which are clustered in an online manner.
TwitterStand (Sankaranarayanan et al., 2009) collects tweets related to breaking
news, cluster them and distribute them to offer an online news delivery service. One
of the tasks performed by TwitterStand is geotaggins tweets. For this task, tweet
text, combined with any articles linked from the tweet, is used, in conjunction with
tweet metadata, such as the location of the user that created the tweet.
Serdyukov et al (2009) present a method for the automatic placing of Flickr photos
on the world map. Their approach estimates a language model by analysing the
textual annotations that are commonly used to describe images taken at particular
locations. An external database of locations is also used in order to improve
efficiency.
Recent works research how supervised learning techniques can be used to increase
the geolocation effectiveness. Wing, B. P., & Baldridge (2011) predict document
locations in the context of geodesic grids of different degrees of resolution.
Lieberman and Samet (2012) propose a method method based on adaptive context
features, which takes into consider a window of context around toponyms to
increase the effectiveness of the toponym resolution task.
Twitter data have been utilized for event detection and summarization from
different domains such as sporting events, earthquakes, elections, etc. However,
there is limited work on the detection and information gathering regarding crime
related incidents.
41
City.Risks
Deliverable D2.1
Most of the works focusing on event detection have been investigating the use of
Twitter data for the detection and summarization of large-scale events of global
significance. Such events are usually related to a large number of relevant tweets.
On the contrary, there has been limited attention to localized events, which may
only be significant to a small subset of the overall Twitter user base. An open
challenge remains the application of prominent event detection approaches to
detect, gather information and summarize localized events, such as events related to
criminal activity.
The lack of structure in tweets does not always allow the efficient extraction of
semantic information. State-of-the-art approaches may be augmented with domain
specific and ontological knowledge in order to accomplish better extraction of
semantic knowledge related to the detection and summarization of criminal and
anomalous activity. To this end, advanced sentiment analysis methods can be
employed, in order to increase criminal and anomalous event detection accuracy by
analyzing the mood of Twitter users, implicitly entailed within tweets.
Data aggregation relies heavily on the specific domain and data. Towards this end,
information mining for crime related incidents and anomalous events must
aggregate information retrieved from Twitter feeds with information that has been
the product of public reporting obtained through other sources, such as relevant
mobile applications. An open question is how techniques for mining text and Twitter
content can be adapted to accommodate public reporting.
Finally, another important issue is the fusion of event-related information from
multiple information formats. Event-related information can be found not only in
Tweets, but also in images and videos. Techniques for the aggregation of social
media content in different formats must be adapted for the purposes of crime and
anomalous event detection.
4.2. Query Processing
This section reviews recent literature on spatial, temporal and textual query
processing and related problems that can be employed or extended within the scope
of City.Risks. Our analysis focuses on techniques from the literature that will provide
the necessary foundation for the data analytics tasks that will be offered by the
platform.
4.2.1. Identifying and ranking locations and areas of interest
One of the core functionalities and services of the City.Risks platform is to collect
and analyse crime related data, in conjunction with other available data (maps,
demographics, etc.), in order to identify, characterise and rank crime “hotspots”.
These refer to locations that are characterised by high rates of one or more types of
criminal activity. From a geographical perspective, they may come in the form of
different shapes, i.e. points (e.g., a single location, such as an ATM), lines (e.g., a dark
alley) or polygons (e.g., a park or a neighbourhood). The system can then use this
42
City.Risks
Deliverable D2.1
information to increase the feeling of safety of citizens by allowing them, for
example, to be informed about and to avoid high risk spots nearby their location or
within an area they wish to visit.
To that end, it is possible to exploit and adapt methods for extracting and ranking
points and areas of interest. The latter is a research direction that has attracted a lot
of interest in the recent years, due to the increasing importance and demand for
location-based services and the widespread use of location-based social networks.
Although studied in a typically different context, e.g. finding attractive places for
tourists or ranking locations offering certain services, the underlying concepts and
techniques can be adapted and extended for extracting and analysing crime
hotspots. Thus, in what follows, we review related work in the topic of identifying
and ranking points of interest (POI) and areas of interest (AOI).
Facing the increasing demand for services providing location search and
recommendations, numerous works have focused on discovering and ranking points
or areas of interest. Various definitions and criteria have been used for this purpose.
The main differences involve the following aspects: (a) whether the focus is on single
POIs or whole areas; (b) whether the problem involves nearby search around a given
query location or rather browsing and exploration within a whole area; (c) whether
the aim is to maximize the number or the total score (e.g. relevance or importance)
of the POIs enclosed in the discovered area or to minimize some cost function (e.g.
distance or travel time) on a set of POIs that suffice for covering the query keywords.
The majority of existing works focus on the ranking of single POIs. In particular,
location-aware top-k text retrieval queries have been studied by Cong et al. (2009).
Given the user location and a set of keywords, this query returns the top-k POIs
ranked according to both their spatial proximity and their textual relevance to the
query. For the efficient evaluation of such queries, a hybrid indexing approach was
proposed, integrating the inverted file for text retrieval and the R-tree for spatial
proximity querying. Further variations of spatio-textual queries and indexes have
been extensively studied (see Chen et al. (2013) for a comprehensive survey). Top-k
spatial keyword queries have also been studied by Rocha-Junior and Nørvåg (2012),
but with distances being calculated on the road network instead of the Euclidean
space. In our case, the keywords appearing in the query could correspond to certain
crime types or other terms found within the description associated to crime
incidents, thus allowing to search for high risk locations with respect to certain types
of crime incidents.
A different perspective for ranking POIs was followed by Cao et al. (2010). In that
setting, the importance of a POI takes into account the presence of other relevant
nearby POIs. Still, the result set of the query is a ranked list of single POIs. This is
particularly useful in our scenarios, since the presence of many nearby crime
hotspots is certainly a factor indicating higher risk.
A different type of queries, involving sets of spatio-textual objects, has been
investigated by Cao et al. (2011) and Zhang et al. (2009). In this setting, the query
specifies a set of keywords, and optionally a user location, and the goal is to identify
sets of POIs that collectively satisfy the query keywords while minimizing the
maximum distance or the sum of distances between each other and to the query.
43
City.Risks
Deliverable D2.1
This type of queries can be used, for example, to find nearby locations that combine
different types of crime incidents, thus identifying potentially interesting patterns
and correlations.
More recently, other works have focused on discovering regions of interest with
respect to a specified category or set of keywords, where the importance of a region
is determined based on the number or the total weight of relevant POIs it contains.
Lamprianidis et al. (2014) have presented a workflow for collecting and integrating
POIs from several Web sources, and then applying density-based clustering to
identify regions with high concentration of POIs of certain categories. A method for
extracting scenic routes from geotagged photos uploaded on sites such as Flickr and
Panoramio has been presented by Alivand and Hochmair (2013). Discovering and
recommending regions of interest based on user-generated data has also been
addressed by Laptev et al. (2014). The quality of a recommended area is determined
based on the portion of the contained POIs that can be visited within a given time
budget. Other variations of queries for discovering interesting regions include the
subject-oriented top-k hot region query, proposed by Liu et al. (2011), and the
maximizing range sum query, proposed by Choi et al. (2014). In these settings, the
region is defined by a rectangle or circle with a maximum size constraint, and the
goal is to maximize the score of the relevant POIs contained in it. Furthermore, Cao
et al. (2014) proposed the length-constrained maximum-sum region query. Given a
set of POIs in an area and a set of keywords, this query computes a region that does
not exceed a given size constraint and that maximizes the score of the contained
POIs that match the query keywords. The query assumes an underlying road
network, in which the POIs are included as additional vertices, and the returned
region has the form of a connected subgraph of this network with arbitrary shape.
The problem is shown to be NP-hard, and approximation algorithms are proposed.
The above works extend and generalise the problem of extracting and ranking POIs
to the case of AOIs. Thus, these techniques can be exploited and adapted for mining
regions (e.g. neighbourhoods) that are characterised by high density of certain types
of crime incidents.
Finally, in a different line of research, other works have applied probabilistic topic
modelling on user-generated spatio-textual data and events to associate urban areas
with topics and patterns of user mobility and behaviour (see Kling and Pozdnoukhov
(2012) and Ferrari et al., 2011). This is an interesting and promising direction
towards relating crime hotspots with information about user mobility, topics and
patterns.
4.2.2. Queries with spatial, temporal and textual filtering
One of the main functionalities of the data management layer of the City.Risks
platform is the efficient evaluation of queries that include spatial, temporal and/or
textual filtering. Several specialized indexes have been proposed in the literature for
such types of queries, typically involving complex, hybrid data structures for
combining two or even all three of the aforementioned dimensions. Next, we review
related work in this area, addressing each case separately.
44
City.Risks
Deliverable D2.1
4.2.2.1. Spatio-temporal queries
Several approaches have been proposed for efficient indexing and querying of
moving objects. A comprehensive survey of spatio-temporal access methods is
provided by Mokbel et al. (2003) and Nguyen-Dinh et al. (2010). Existing indexes are
categorized according to whether they index past, current or future positions of
moving objects (or combining all three).
Here, we mainly focus on the first category, namely, indexing the past positions of
moving objects. In the City.Risks platform this can be used, for example, to find theft
detection sensors that have been located within a certain area during a specified
time interval.
One of the main approaches in this category is SETI (Chakka et al. (2003)). SETI
employs a two-level index structure to handle the spatial and the temporal
dimensions. The spatial dimension is partitioned into static, non-overlapping
partitions. Then, for each partition, a sparse index is built on the temporal
dimension. Thus, one main advantage of SETI is that it can be built on top of an
existing spatial index, such as an R-tree. Furthermore, an in-memory structure is
used to speed up insertions. Queries are evaluated by first performing spatial
filtering and then temporal filtering. That is, first, the candidate cells, i.e. those
overlapping with the spatial range in the query, are selected. Then, for each cell, the
temporal index is used to retrieve those disk pages whose timespans overlap with
the temporal range in the query. Query execution concludes with a refinement step,
to filter out candidates, and (if trajectories are desired as the output) a duplicate
elimination step, to filter out segments that belong to the same trajectory. A similar
approach is followed also by the MTSB-tree (Zhou et al. (2005)) and the CSE-tree
(Wang et al. (2008)), which, as in SETI, partition the space into disjoint cells but differ
in the type of temporal index maintained for each cell.
An alternative approach is followed by the PA-tree (Ni and Ravishankar (2005)),
which instead divides first the temporal dimension into disjoint time intervals. Then,
the trajectory of each object is split into a series of segments, according to these
time intervals. Each segment is approximated with a single continuous Chebyshev
polynomial and a two-level index is used to index these approximated trajectory
segments within each time interval.
In a different direction, spatio-temporal indexes have also been proposed for
indexing objects moving in a fixed network (Frentzos (2003), de Almeida and Güting
(2005) and Le and Nickerson (2008)) or in symbolic indoor spaces (Jensen et al.,
2009).
4.2.2.2. Spatio-textual queries
Spatio-textual queries have received a lot of attention over the past years due to the
increasing interest and use of location-based services and social networks. These
involve keyword-based search for points of interest and retrieval of geotagged
documents, web pages, photos, microblogs, news, etc. The main focus has been on
combining spatial and text indexes to efficiently support queries involving both
45
City.Risks
Deliverable D2.1
criteria. A comprehensive survey and comparison of existing approaches is provided
by Chen et al. (2013).
Essentially, the indexes proposed by these approaches are hybrid structures
comprising a spatial indexing part and a text indexing part. The former is typically
based on an R-tree, grid or space filling curve, while the latter can be an inverted file
or a bitmap. Thus, several indexes have been proposed employing different
combinations of these options. A main difference is also whether the combination
follows a text-first or a spatial-first approach, as investigated by Christoforaki et al.
(2011). In the first case, for example, the top-level index can be an inverted file, in
which the postings in each inverted list are indexed by an R-tree; instead, in the
second case, the top-level index can be an R-tree, with inverted files attached to
each leaf node. Characteristic examples are the IF-R*-tree and the R*-tree-IF (Zhou
et al. (2005)). A further difference is how loosely or tightly the two structures are
combined. For example, another index structure that is based on the R-tree and
inverted files, but combines them more tightly, is the KR*-tree (Hariharan et al.
(2007)). The KR*-tree maintains an inverted index-like structure, called KR*-tree List,
that, for each keyword, keeps a list of nodes in the R*-tree that have the keyword.
Each leaf node additionally contains inverted lists that index the keywords appearing
in the objects under the node.
Several similar variants exist (e.g. the IR-tree by Wu et al. (2008)). Inverted files have
also been combined with grid (Vaid et al. (2005)) and space filling curves (Chen et al.
(2006) and Christoforaki et al. (2011)).
In the City.Risks platform, investigating such types of hybrid index structures is an
interesting direction for efficiently evaluating queries that involve, for example,
finding nearby locations where crime incidents of certain types or characterized by
certain terms have occurred.
4.2.2.3. Spatio-temporal-textual queries
Finally, driven by the growing amount of Web data containing both timestamps and
spatial footprints, the need for combining keyword search with spatial and temporal
filtering has been recognized. Nepomnyachiy et al. (2014) have proposed an index
that is based on a shallow R-tree, combined with an inverted index at each leaf node
to index the terms of the contained documents. In addition, to deal with the
temporal dimension, the original document ids are replaced with new ids that are
assigned to documents chronologically, thus facilitating the retrieval of documents
within a given temporal range.
Keyword search on trajectories has been studied by Cong et al. (2012). Each
trajectory consists of a sequence of geospatial locations associated with textual
descriptions. Then, given a location and a set of keywords, the goal is to find the topk trajectories whose text descriptions cover the given keywords and have the
minimum distance to the given location. The proposed method is based on a hybrid
index, called cell-keyword conscious B+-tree, which enables simultaneous
application of both spatial proximity and keyword matching.
46
City.Risks
Deliverable D2.1
In a different direction, the problem of continuously moving top-k queries over
spatio-textual data has been addressed by Wu et al. (2013). The focus is on
computing safe zones that guarantee correct results at any time and that aim to
optimize the server-side computation as well as the communication between the
server and the client. Finally, the query studied by Skovsgaard et al. (2014) returns
the top-k most frequent terms in a given spatio-temporal range. The proposed
technique is based on extending existing frequent item counting techniques to
adaptively maintain the most frequent items at various spatial and temporal
granularities.
In the City.Risks platform, it is an interesting and challenging research direction to
exploit, adapt and extend such index structures to efficiently manage, search and
analyse crime related data and other information involving spatial, temporal and
textual attributes. The aim of this process is on the one hand to efficiently utilise
state-of-the-art indexes in order to allow the efficient treatment of complex queries
on crime data that will be useful to the scope of the City.Risks project, while on the
other hand to effectively adopt these methods based on the particular structure of
crime related data and queries in order to further increase their efficiency in
answering these particular queries.
4.3. Privacy and Anonymization
4.3.1. Privacy issues and anonymization techniques
With the increasingly widespread use of GPS enabled devices and other positioning
technologies, it has become possible for a variety of applications to track the
movement of various types of objects, including vehicles, animals and humans. This
opens up new capabilities and opportunities for analysing the behaviour of those
entities and extracting useful knowledge and patters.
In City.Risks, tracking the location/and or movement of users is relevant for several
envisaged services of the platform, such as providing warnings and alerts about
nearby crime hotspots or incidents, recommending safe routes, finding witnesses for
an ongoing or past event, etc. However, at the same time, significant privacy
concerns are raised, since movement tracking may reveal sensitive information
about individuals. In what follows, we describe the problem and we provide an
overview of the state-of-the-art in anonymization techniques for movement tracking
data. This will offer a basis for addressing any privacy issues concerning such kind of
data to be used in the City.Risks platform.
The issue of data privacy is not specific to location tracking data; in fact, it comes up
and has been investigated in several domains that involve publishing microdata, such
as customer data or health records. On the one hand, publishing such data is desired
and highly valuable since it offers the potential to extract valuable knowledge; on
47
City.Risks
Deliverable D2.1
the other hand, it entails the risk of privacy breach by accidentally revealing sensitive
information. Simply removing explicit identifiers, such as names and social security
numbers, is not sufficient. A characteristic example was presented by Sweeney
(2002), where the medical records of the governor of Massachusetts were revealed
by combining gender, date of birth, and 5-digit zip code information from voter
registration records and a previously de-identified dataset with medical records
published by an insurance company. This shows that combining a set of attributes,
referred to as “quasi-identifiers”, from a de-identified dataset with other publicly
available data can still reveal sensitive personal information which was initially to be
protected.
To address this problem, the concept of k-anonymity has been proposed (Samarati,
2001; Sweeney, 2002). This is a data privacy guarantee, requiring that for each
individual there exist at least k-1 other individuals having the same values for the
attributes recognised as quasi-identifiers. In other words, this guarantees that each
individual is “hidden” among k-1 others. Nevertheless, a privacy breach still exists if
all k individuals within such a group have the same value for the sensitive attribute
(e.g. the same disease). In that case, the sensitive information is still disclosed, even
though the exact individual cannot be identified. To deal with this, a further
guarantee, referred to as l-diversity, has been proposed (Machanavajjhala et al,
2007; Xiao et al. 2010). This guarantee requires that within the same group of
individuals there should be at least l different values for the sensitive attribute.
Although these concepts and techniques are also relevant when dealing with
location and movement data, they are not sufficient to protect privacy. When a
series of locations is available for a user, exploiting spatial and temporal correlations
may still reveal sensitive information, even though partial anonymity guarantees
may have been applied. Thus, a specific part of research has focused on privacy
protection for this specific type of data.
In particular, two broad scenarios can be identified. The first involves continuous
location-based services, where a mobile user reports its location –periodically or ondemand– in order to receive certain information or services. The second involves
privacy protection for historical trajectories. Next, we briefly present the main
techniques that have been proposed for each case (for a more detailed survey, see
Chow and Mokbel, 2011).
For the first category, the main techniques that have been proposed in the literature
include spatial cloaking (Mokbel et al, 2006), mix zones (Beresford and Stajano,
2003), path confusion (Hoh et al., 2010) and dummy trajectories (Kido et al., 2005).
Spatial cloaking is a technique that was also proposed for snapshot location-based
services. The main idea in this approach is to “hide” each user in a “cloaked” spatial
region that guarantees k-anonymity. The approach is extended in the case of
continuous location-based services, where two main techniques can be further
distinguished, namely group-based and distortion-based. The basic idea is that,
before issuing a query, a user forms a group with k-1 other nearby users, and the
spatial region containing this group is used instead of the specific location of that
user; then, at subsequent queries, the cloaked regions is updated to still include the
group members. The distortion-based approach improves upon the group-based one
48
City.Risks
Deliverable D2.1
by considering also movement direction and velocity in order to optimise the
selection of the cloaked spatial regions.
Mix zones are spatial regions, for which the following apply: (a) when a user enters a
mix zone, it is assigned a new identifier, and (b) while the user remains in the mix
zone, it does not send any location information. By appropriately selecting the mix
zones, it is guaranteed that a user that exits from a mix zone cannot be distinguished
from any other user that was also within the same mix zone at that time.
Path confusion is a technique aimed at avoiding the linking of consecutive locations
to individual users by attacks referred to as target tracking (Gruteser and Hoh, 2005).
Finally, in the dummy trajectories approach, the idea is to generate and include
some fake location trajectories in order to hide the real trajectory of the user.
For the second category, involving privacy protection in historical trajectories, two
main types of techniques can be identified, clustering-based (Abul et al., 2008) and
generalisation-based (Nergiz et al., 2009). Compared to the case of continuous
location-based services described above, the main difference here is that the
complete trajectory of each individual user or object is known in advance. In the
clustering-based approach, the basic idea is to cluster together k trajectories that are
both spatially and temporally close in order to form an anonymized aggregate
trajectory. In the generalization-based approach, the algorithm works in two steps.
In the first step (anonymization), each initial trajectory is anonymized as a sequence
of k-anonymized regions. Then, in the second step (reconstruction), new trajectories
are reconstructed from the generalized ones by uniformly selecting k points for each
anonymized region and then using one of them to form a trajectory linking the
respective regions.
During the design of the City.Risks platform and services, we will investigate how
such techniques can be exploited and adapted to provide privacy protection when
dealing with location and movement data. Nevertheless, our topmost priority will be
to alleviate, as much as possible, such issues. To that end, we will leverage the
increased capabilities of modern smart phones in order to perform –whenever
possible- location-aware processing tasks on the mobile device instead of sending
location data for processing on the server. Moreover, emphasis will be given on such
issues when designing the mobile applications and services, making sure that the
user is explicitly informed and can choose which information is disclosed, when and
with whom. For example, it may be possible for many scenarios to not transmit and
store the location of a user to the operation centre, but instead share it directly only
within a specified trusted network of other users, such as family or friends.
4.4. Data Analytics
The City.Risks platform involves the collection and management of large volumes of
data related to crime incidents and more general purpose information describing the
49
City.Risks
Deliverable D2.1
urban areas of focus. All this information is crucial to support the decision making
tasks that must be supported by the City.Risks platform.
In order to make informed decisions, the City.Risks platform must promote the
efficient communication of the information maintained within the City.Risks
databases. To achieve this, the platform must be able to export selected descriptive
statistics summarising the data, thus allowing the decision makers to exploit this
information. On top of this, the City.Risks data may be exploited for predictive
analysis. Such a data-drive approach will allow the exploitation of historical data, in
order to identify hidden patterns, which will provide insights about future events.
4.4.1. Mapping crime
Spatial location is highlighted as an important characteristic related to crime
incidents (Chainey et al., 2008, Chainey and Ratcliffe, 2013). Concentration of crime
incidents to certain locations is usually related to particular characteristics and
opportunities associated with these spatial locations.
Areas with high concentration of crime incidents are referred to as crime hotspots
(Chainey and Ratcliffe, 2013). Mapping crime incident clusters is a vital operation for
a variety of tasks executed by law enforcement agencies. Decision making on the
placement of patrols for instance is a crucial task that is considered to benefit from
crime hotspot mapping.
The advancement of GIS software has provided important tools for crime mapping
techniques (Chainey et al., 2008). The simplest form of crime mapping involves
point-mapping, where crime incidents are visualized in the form of points on a map.
Another early example of crime mapping methods is the Spatial and Temporal
Analysis of Crime applications, which identifies crime clusters that are then fit to a
standard deviational ellipse.
Choropleth Mapping aggregates and visualizes crime incidents, with respect to a
predefined space partitioning. Usually, space partitioning is based on administrative
area boundaries. For instance, the different areas may represent wards or districts.
In this case, different shades of color describe the crime rate in different
administrative regions. An alternative is to divide space in cells of equal size. In this
case, a spatial grid is utilized in order to divide the area, and represent aggregated
crime volume by different shades of the cells of the grid.
According to (Eck et al., 2005, Chainey et al., 2008, Chainey and Ratcliffe, 2013), one
of the most suitable techniques for crime mapping is Kernel Density Estimation
(KDE). KDE is considered to offer high accuracy hotspot detection in conjunction with
high quality aesthetics. KDE fits a spatial probability density function to historical
crime incident records. KDE hotspot mapping results in the visualization of smooth
surfaces describing crime densities.
50
City.Risks
Deliverable D2.1
According to Ratcliffe (2010), spatio-temporal crime patterns constitute an area of
criminology that remains under-researched. The importance of the temporal
elements of crime is indicated by repeat and near repeat victimization phenomena.
For instance, after a burglary, not only the same but also nearby houses have a
higher risk of been targeted, for a time period ranging from several weeks to
months.
4.4.2. Predicting crime
(Chainey et al., 2008) investigates the predictive capabilities offered by the
aforementioned techniques. More specifically, their work researches whether, apart
from mapping and aggregating historical crime incidents, the discussed techniques
have additional predictive capabilities. The results show that KDE produced superior
results compared to other evaluated techniques in predicting future spatial patterns
of crime. Another outcome of this work is that crime type is an important factor in
the prediction of spatial patterns. In particular the predictions for street level crime
were superior to those for any other crime type. This was attributed to the
robustness of street level crime opportunities.
Mohler et al. (2012) employs self-exciting point processes to model space-time
clustering of crime data. This work is influenced by the use of self-exciting point
processes to model clustering patterns in earthquakes by seismologists, and the
perceived similarity with criminal events in terms of local and contagious spread. The
proposed method is implemented with respect to residential burglary data.
Hotspot maps describe locally relevant information. Inferences made using these
maps are only relevant with respect to areas associated with historical information.
As a result, the knowledge displayed in these maps is local, and cannot be exploited
for different areas. Xue and Brown (2006) employ multiple spatial and demographic
features to train a predictive model that can be used both in areas with no historical
data. Their approach, instead of using exclusively the spatial coordinates, utilises
features such as distances from local landmarks, such as nearest business, or
demographic information such as the number of divorced individuals in the region.
4.4.3. Crime analytics using the social media
Recent work has focused on using the social media as a source of information that
can be exploited to provide enhanced data analysis related to crime incidents. The
social media offer a source of large volumes of information, which in many cases can
be exploited in a real time fashion. As a result, this alternative feed of information
may be used in conjunction with historical data in order increase the prediction
accuracy of a system.
Wang et al. (2012) investigate the use of Twitter messages as a source for prediction
of criminal incidents. Their approach is based on automated analysis of tweets, in
conjunction with dimensionality reduction based on latent Dirichlet allocation, and
prediction based on linear modelling. The proposed methods are evaluated on the
51
City.Risks
Deliverable D2.1
prediction of hit-and-run crimes. The results show that their method outperforms a
baseline method that uniformly distributes incidents across days.
As already mentioned earlier, Gerber (2014) examines the exploitation of tweets
with spatio-temporal attributes for the purposes of crime prediction. The proposed
method performs topic modelling in order to identify trending topics in a major
urban area. These topics are used in conjunction with twitter-derived features in
order to predict crime incidents. 25 crime types were studies. For 19 of these crime
types, the proposed methods achieved higher predictive performance than a
standard KDE approach.
4.4.4. Crime analysis software
While there is a plethora of algorithms for specific crime related data analysis tasks,
it is important to review the capabilities offered by a comprehensive system used by
crime analysts. This is of great relevance to City.Risks, since it is vital that the core
usability that is related to the tasks at hand is reproduced or extended.
CrimeStat (Levine 2015) is a free, widely-used, comprehensive spatial statistics
software system for analysis of crime related incidents. CrimeStat includes a plethora
of statistical methods for the analysis of crime data.
Spatial description algorithms that are supported by the system include statistics for
describing the spatial distribution of incidents, spatial autocorrelation and distance
analysis. CrimeStat also includes methods for hotspot analysis based on nearest
neighbours, STAC and k-means, and hotspot analysis of zones.
CrimeStat also includes a series of spatial modeling algorithms. Interpolation is
performed by a single-variable kernel density estimation method and a head Bang
routine for smoothing zonal date. Space-time clustering and analysis is also
supported. In addition, CrimeStat supports journey to crime analysis, which uses
crime incidents data to estimate the likely location of a serial offender. The software
additionally offers multiple modules for regression modeling and discrete choice
modeling. CrimeStat also provides a time series module for crime forecasting. Finally,
CrimeStat includes methods for Crime Time Demand Modelling.
The City.Risks platform must offer a comprehensive analytics module that allows the
level of analysis described in previous sections. In order to do so, this module must
rely on the different information sources that will be exploited by City.Risks. As a
result, we will investigate how information from diverse sources such as police,
demographics, road network data, news articles, and social media can be fused and
analysed in order to produce actionable information.
In addition, we will focus on the further utilisation of real time and noisy sources of
information, such as Twitter, in conjunction with historical data, for predictive
52
City.Risks
Deliverable D2.1
analytics tasks related to safety and the fear of crime. For instance, one direction is a
classifier able to predict the type of an incompletely reported crime.
Depending on the size of the data sets obtained for City.Risks, traditional methods
may not be applicable. To this end, we will investigate alternative implementations
of the necessary crime data analytics algorithms that can leverage the state-of-theart distributed programming models and frameworks (Dean and Ghemawat, 2008,
Zaharia et al., 2010).
4.5. Route Planning
One of the requirements that must be satisfied by the City.Risks platform is the
efficient treatment of routing queries in road networks. Such queries are the
foundation of multiple operations of the system, such as organising evacuation
routes, aiding decisions on patrols and identifying reporters near areas of interest.
There have been important recent advances for the efficient treatment of routing
queries in road networks. Particular attention has been paid to the problem of
identifying shortest paths, which is the core operation behind every route planning
task. In addition, there is a plethora of works focusing on more practical
transportation problems introducing additional constraints and goals that must be
satisfied.
The City.Risks platform operates in an urban environment, and involves actors, such
as operational units or reporters, that must be positioned and operate in order to
alleviate potential threats. An efficient routing module is crucial for the treatment of
such operations. This module, apart from providing the routing mechanism, must be
capable of responding to domain specific queries related to City.Risks. This section
reviews the literature on route planning and related problems and discusses gaps
and challenges and research directions of relevance to City.Risks.
4.5.1. Shortest path queries in road networks
The core focus of route planning research is centred around the efficient treatment
of shortest path queries. Such queries are the core operations behind transportation
problems in road networks. Modern techniques achieve significant speedups in
calculation times, and to do so they rely on the exploitation of the fundamental
structure of road networks in combination with advanced pre-processing techniques.
A detailed overview of the state-of-the-art can be found in (Bast et al., 2014).
The road network is usually maintained in the form of a directed graph. Vertices
represent junctions, whereas arcs represent road segments. Arcs are associated with
metric weights that represent some form of traversal cost. These weights can be
static or dynamic. They usually correspond to traversal times, arc distances, energy
costs, or the combination of multiple metrics.
53
City.Risks
Deliverable D2.1
Given a graph representation of a road network, route planning is performed by
computing the shortest path between a source and a target node in the given graph.
This task can be achieved by classical search algorithms, such as the well-known
Dijkstra algorithm. Alternatively, directed A* search can be employed using the
spatial distance between a vertex and the source vertex in order to compute a
bound on the actual shortest path distance.
Classical search algorithms can accurately identify shortest paths in graphs, but do
not scale to real-world route planning settings. Route planning is not a trivial
problem, especially due to the size of real-word transportation networks. Consider
for instance that the road network for Western Europe consists of 18,000,000
vertices.
Modern techniques build on top of classical search algorithms in two directions. First
of all, they rely on heavy preprocessing in order to shift parts of the computation offline, thus allowing better on-line response times. Secondly, the algorithms utilize the
inherent structure (hierarchical and almost planar) of road networks in order to
direct the search.
The preprocessing stage usually relies on assumptions on the dynamicity of the road
network. As a result, depending on the particular problem at hand, a different
technique may be preferable, since the different methods have different capabilities
with respect to changes in the road network connectivity or the arcs. In addition, the
size of the network and the storage constraints impose additional requirements on
the level of preprocessing. Usually, the best case falls between a purely on-line
search and the complete pre-computation of the shortest path distance between all
vertices and its storage in the form of a distance matrix.
The use of spatial coordinates to direct the A* search has not proven to yield
efficient results. A better alternative is ALT (Goldberg and Harrelson, 2005), an
algorithm based on A*, Landmarks and the Triangle Inequality. During preprocessing, a small subset of vertices is selected as Landmarks (usually 20 vertices
distributed among the graph are selected), and the distances between Landmarks
and every vertex in the graph is calculated. At query time, the triangle inequality and
the landmark distances are used to calculate a lower bound on the distance between
a vertex reached by the A* search and the target vertex.
Another approach is the Arc Flags (Hilger et al., 2009) algorithm, which precomputes information that can be used at query time in order to filter the amount of
edges that need to be considered during the search. More specifically, in the preprocessing phase, the algorithm labels every arc with the vertices that are part of a
shortest path starting from this arc. At query time, only edges that are labelled with
the target vertex in the relevant flag set need to be examined. In order to reduce
pre-computation time and the required space to store the flags, a graph partitioning
approach is followed. The partitioning process breaks the graph in k subsets, while
minimizing the number of border nodes between partitions. Then, arcs are labelled
with a partition if this partition contains at least one vertex that lies in a shortest
path starting from this arc.
54
City.Risks
Deliverable D2.1
One of the most efficient modern approaches is the Contraction Hierarchies
algorithm (Geisberger et al., 2012), which leverages the hierarchical structure of the
road networks. The pre-processing step initially orders the vertices in the graph
according to their significance. This is a heuristic metric that usually combines
multiple criteria that offer insight on the importance of a vertex with respect to
routing on the given graph. Then, according to this ordering the nodes of the graph
are contracted. Every contraction step is followed by the addition of the necessary
shortcuts. Shortcuts are the arcs that need to be introduced in order to maintain all
shortest path distances after a contraction has taken place. The pre-processing
phase results in the vertex ordering and a set of shortcuts. These are used at query
time to speed up the search for a shortest path. A bidirectional version of the
Dijkstra algorithm is used to search a graph that contains the original arcs and the
shortcuts. The addition of shortcuts asserts that both strands of the bi-directional
Dijkstra search need to focus exclusively on traversal of vertices with increasing
significance. The two search strands intersect at the most significant vertex in the
path.
An alternative approach is based on the observation that most distant enough
shortest paths pass through the same limited set of vertices. Transit-Node Routing
(Bast et al., 2007) identifies such Transit Nodes and pre-processes the distances
among these vertices and every other vertex in the graph. Using this information, a
“distant enough” shortest path query is computed by identifying the transit nodes
that minimize the total distance source-tsource-ttarget-target, where tsource, ttarget are
transit nodes for source and target vertices respectively.
Hub Labelling (Abraham et al., 2012, Akiba et al., 2013) is one of the most prominent
modern approaches. This method selects, for every vertex in the graph, a set of
vertices called its hubs. The distances between a vertex and its hub vertices are precomputed. Hub selection must obey the cover property. The intersection of the hubs
of every pair of vertices u, v must contain at least one vertex that falls on the
shortest path from u to v. The shortest path distance between u and v is found as
the minimum distance u-h and h-v, for any common hub h of u and v. Even though,
in the worst case, hub label sizes are impractical, the inherent structure of road
networks in combination with efficient heuristics allow significant reductions on the
size of the hub label sets.
Most of the methods discussed so far can be combined in order to achieve even
higher speedups compared to the standard Dijkstra algorithm. For instance Arc Flags
have been used in conjunction with contraction hierarchies or Transit-Node Routing.
4.5.2. Generalised path queries
Shortest path queries are the basis for route planning in road networks. However,
real-world problems usually impose additional constraints that exceed the
computation of the shortest path distance between vertices in a graph. Current
research investigates how relaxations on the assumptions of the standard problem
and the introduction of additional constraints modify the shortest path problem, and
55
City.Risks
Deliverable D2.1
examines how proven shortest path algorithms can be used or extended in order to
efficiently treat generalised queries.
Therefore, given a novel generalised shortest path problem, such as the problems
that need to be addressed within the City.Risks platform, the questions that need to
be addressed are (i) what are the additional constraints that need to be addressed,
(ii) which are the most suitable state-of-the-art methods for the given problems, and
(iii) which modifications must be made to the selected techniques in order to
efficiently treat these problems.
One scenario that has been studied in the literature involves the efficient
computation of shortest path queries in mobile or GPS devices. This is vital in
situation in which mobile communication is impossible or interrupted. The main
constraint in these cases is the lack of storage space and the limited computational
capabilities of such devices. Research has shown that algorithms based on ALT
(Goldberg and Werneck, 2005) or Contraction Hierarchies (Sanders et al., 2008) can
be successfully adapted to suit this scenario.
In many real-world scenarios the shortest path is not necessarily the best. Depending
on the scenario, the quality of a route may be measured by multiple criteria. The
different criteria may or may not be completely independent, and it may or may not
be possible to combine them to form a single cost function. For instance, with
respect to the City.Risks domain, the estimated traversal time may be less important
factor than a safety factor, calculated with respect to the crime events that have
been observed in an area. Depending onthe task at hand, we may seek to provide
the route with the minimum risk, or combine the two factors in a metric that is both
time and risk aware. In general, a flexible objective function makes the task of preprocessing harder.
TREADS (Fu et al., 2014) is a route planning recommender system that combines
three different metrics: path length, safety and the number of points-of-interest
traversed by the path. The three metrics are used independently or combined by a
weighted linear function in order to provide an aggregated metric. Safety related
incidents are mined from transportation related topic models from Twitter. These
are then used to calculate the safety score of a path.
In order to account for tasks that involve different metrics that cannot be combined
in a single function, route skyline queries (Kriegel et al. 2010) have been proposed.
Such queries seek routes that are optimal with respect to an arbitrary combination
of multiple criteria. For instance, given the criteria of safety and traversal time, the
skyline operator computes all routes are optimal with respect to a combination of
these criteria. In particular, a route is in the result set of the operator if there does
not exist an alternative that is better with respect to both criteria.
Many real-world scenarios involve dynamic information that is updated constantly.
On standard route planning scenarios a source of dynamicity is traffic. The City.Risks
domain involves a series of dynamic scenarios, especially with respect to large-scale
crime related events. In such cases, it may be unreasonable to expect that the
information regarding how safe an area is, or the time it would take to traverse a
56
City.Risks
Deliverable D2.1
path is static. On the contrary, such information should be updated in order to
reflect the latest information that has been shared by the ground reporters.
The presented state-of-the-art route planning methods shift a significant amount of
computation to the pre-processing step. This is not always effective in dynamic
domains, since a large portion of the pre-processed information may no longer
reflect the environment. In addition, the pre-processing step is usually time
consuming and cannot be repeated at run time whenever new information comes to
light. In order to tackle this problem several approaches have been proposed. On
approach is to modify the pre-processed data in order to reflect the changes in the
actual domain. Examples of this approach have been proposed for most of the
techniques presented above (Delling et al. 2007). An alternative approach is to shift
the burden of dealing with the new information that is not reflected by the preprocessed data to the search methods. This approach has been used for A* search
(D’Angelo et al., 2012) with Landmarks and Contraction Hierarchies (Geisberger et
al., 2012). Finally, a different way to deal with dynamic information is to divide preprocessing into two stages. The first operates on the graph topology, and the second
on the cost metric. This allows the re-use of the results of the first step when
information updates involve the cost function associated with the graph. This
approach has been followed in combination with multiple techniques such as ALT
(Delling et al. 2007) Contraction Hierarchies (Geisberger et al., 2012).
4.5.3. Generalised routing problems
The City.Risks domain involves scenarios that exceed the assumptions of the
problems discussed so far. Many interesting situations do not simply involve a single
source and a single target vertex. For example, consider a scenario involving multiple
ground reporters situated in different locations of a city, and a number of events
that are taking place simultaneously in the urban area for which information is
required. The task of assigning each reporter with a set of places that they need to
provide information for exceeds a shortest path calculation query. This problem
requires multiple such operations in order to compute the best assignment. Such
problems usually involve combinatorial optimisation.
One of the most common relevant problems is the Travelling Salesman Problem
(TSP). TSP requires the calculation of the shortest path that traverses though a
number of predefined locations, and returns to a predefined source vertex. The
Travelling Salesman Problem has found a plethora of applications in real-world
scenarios. This problem has been thoroughly researched both with respect to finding
exact solutions (such as algorithms based on branch and bound techniques) and
approximations (such as meta-heuristic approaches).
Recently, TSP has been extended to deal with satisfying a number of categorical
locations in road networks. In this case, instead of searching for paths that traverse a
number of predefined locations, the search seeks optimal paths that contain vertices
that satisfy a number of given location classes (Rice and Tsotras, 2012, Rice and
Tsotras, 2013) or textual descriptions (Yao et al., 2011). For example, this process
involves the search for a path, given a source and a destination vertex, which passes
57
City.Risks
Deliverable D2.1
through any bank and gas station facility. These approaches operate on road
networks and utilise state-of-the art techniques, such as Contraction Hierarchies and
ALT, in order to efficiently compute search for optimal paths.
The Orienteering Problem, is somehow similar to TSP, but seeks the path that
maximizes the total utility provided by the traversed vertices and at the same time
does not exceed a predefined cost budget. With respect to the City.Risks domain, a
relevant application would be the computation of a path that traverses the maximal
amount of areas that need to be reported. A recent survey on the orienteering
problem is provided by Vansteenwegen et al. (2010).
Vehicle-Routing Problem (VRP) is a generalisation of TSP. This problem involves
multiple vehicles and seeks the optimal paths that allow these vehicles to service a
set of customers in predefined locations. Several extensions to the problem have
been proposed including capacity constraints, heterogeneous fleet and dynamic
information. Pillac et al. (2013) survey the most important recent approaches in
dynamic vehicle routing.
A different set of problems that are also relevant for City.Risks are the problems that
deals with optimal positioning of facilities or mobile objects on a map, in order to
allow to better service an area or achieve better response times to emergencies.
Instances of such problems are the Location Set Covering Problem and the Maximal
Covering Location Problem. Surveys of the most important works can be found in (Li
et al., 2011, Farahani et al., 2012).
Evacuation transportation modeling (Murray-Tuite and Wolshon, 2013) is a heavily
studied research field. The goal of this research is to provide efficient systems that
can aid the management of natural disasters and other emergencies. Evacuation
Route Planning is a particular direction of this work that is relevant with respect to
the City.Risks platform. Hamacher and Tjandra (2001) divide research in macroscopic
and microscopic evacuation models. Macroscopic evacuation models (Kim et al.,
2007, Lim et al., 2012) are based on dynamic network flow models representing the
aggregated behaviors of the evacuees. On the contrary microscopic models (Richter
et al., 2013) operate on the level of individual behaviors of evacuees and their
interactions.
Dynamic Ride-Sharing seeks to match users with similar itineraries in order to
minimize the overall transportation costs. As a result, such systems allow users to
optimize the utilisation of the available seat capacity, and ultimately improve the
efficiency of the overall transportation system. (Agatz et al., 2012, Furuhata et al.
2013) provide overviews of work in dynamic ride-sharing. City.Risks can adopt such
methods and apply them towards matching user itineraries, and facilitating group
transportation in unsafe areas, in order to increase the overall sense of safety of the
users.
The City.Risks platform needs to tackle a series of problems related to route planning
in road networks. In the previous sections, we discussed the state-of-the-art for
58
City.Risks
Deliverable D2.1
these tasks. The research conducted within the City.Risks project will be based on
these works in order to create efficient methods, tailor-made to particular problems
of the project.
First of all, more focus is needed on the development of efficient, safety-aware
routing algorithms. To this end, efficient state-of-the-art techniques need to be
adapted and extended by introducing additional cost metrics and constraints
associated with the crime data that will be available within the City.Risks platform.
Second, the algorithms that will be constructed for the computation of safety-aware
paths will be the basis for generalised safety-aware routing algorithms. A concrete
example of such methods is an algorithm able to solve the safety-aware routesharing task. This task would allow users of the City.Risks platform to come together
while traversing unsafe urban areas.
Finally, the City.Risks platform must provide tools for the management of ground
reporters in light of security events. This component will be based on the adaptation
of problems related to location allocation.
59
City.Risks
Deliverable D2.1
5. Mobile Sensors and Communications
An important part of the City.Risks project is based on the efficient use of sensors
and radio-based technologies to transparently identify and locate stolen objects in
an urban environment based on a network of citizens. This section presents the
state-of-the-art research in mobile sensors and communications and outlines
important research directions that will be exploited within the scope of City.Risks.
5.1. Sensors and Communication
The technology assisted physical and social interactions have given birth to the idea
of ubiquitous computing (Yeon, 2007). This vision has been accelerated by the
developments in computing power, prolonged battery life, open software
architectures and cost-effective wireless network technologies. Wireless Sensor
Networks (WSN) is an essential part of this form of computing (Garcia et al, 2007;
Yick, Mukherjee and Ghosal, 2008; Kuorilehto, Hännikäinen, and Hämäläinen, 2005;
Sharifi and Okhovvat, 2012). WSN usually consists of a plethora of nodes (small
sensors) with sensing and communicating capabilities. These sensors have limited
processing or memory capacity due to economic and self-power constraints.
Wireless Sensor Networks are used for monitoring in various areas (Bae et al, 2011;
ITU, 2005) by constructing wireless networks with cooperation and self-organizing
features. Structural health monitoring, intrusion detection, and critical environment
surveillance were some of the first fields to incorporate the use of WSNs. However,
one of the novel areas of focus has been the monitoring systems and convergence
technologies so as to ensure social and national security and natural disaster
surveillance. Modern WSN monitoring systems have low power and short range
nodes with self-developed or standard communication technologies (Wheeler, 2007;
Ergen and ZigBee/IEEE, 2004; Texas Instruments, 2007; ZigBee Alliance). Finally,
scientific efforts focus on global network communication technologies such as IPWSN (IP-based wireless sensor network) (Ha et al, 2010; Hong et al, 2010).
5.1.1. Communication technologies for wireless sensor networks
WSN generally consists of sensor fields with sensor nodes and a sink node. To
achieve WSN, sensor nodes communicate among themselves into a sensor field and
with the sink node. Their communication technologies mainly adopt WiFi, HSDPA,
Wibro, TRS, ZigBee, UWB, Bluetooth, and 6LoWPAN. Their features are briefly
described as follows (more details can be found on Kim et al. (2013)).
•
•
WiFi is a wireless LAN standard (IEEE 802.11b), offering high-speed wireless
Internet using Access Points.
High Speed Downlink Packet Access (HSDPA) is an asynchronous communication
protocol for mobile telephone data transmissions. It offers theoretical data
transmission speeds reaching 14 Mbps, but which in practice to 2~3 Mbps.
60
City.Risks
•
•
•
•
•
•
Deliverable D2.1
Moving devises may operate on speeds over 100km/h. The head count per base
station is limited.
Wibro (wireless broadband Internet) is IEEE 802.16e standard. It corresponds to
3.5G wireless communication technology, and Wibro-Advanced to 4G
technology. Wibro has the advantages of ubiquitousness. Base stations offer
collected data throughput of 30~50 Mbit/s per carrier. Also, they cover a radius
of 1~5 km. Finally, moving devices may operate on high speeds up to 120 km/h.
TRS (trunked radio system) combines mobility and two-way radio. In addition, it
allows multiple users to operate on a single frequency. TRS has been heavily used
in police and fire operations.
ZigBee is a low-cost and low-power wireless mesh network standard built upon
PHY and MAC layers defined in IEEE 802.15.4. ZigBee uses a communication
frequency of 800 MHz~2.4 GHz and achieves a date rate of 20 Kbps~250 Kbps.
ZigBee does not directly support IP connectivity.
UWB is employed for communication and for radar based applications.
Frequencies can be shared without interference. UWB achieves high-speed
communication of low power across very large areas reaching speeds over
100 Mbps (3.1~10.6 GHz). However, due to partial pulse phase, communication
requires intensive time synchronization.
Bluetooth offers low power, low price, short distance wireless communication.
Its disadvantages are slow transfer rate and small communication radius is short
(10 m).
6LoWPAN adopts IEEE 802.15.4 and links sensor networks and IPv6 networks. Its
application domains include global monitoring and cloud computing. IP-enabled
sensors allow the decoding and analysis of information between different
networks and communication protocols.
5.1.2. Network topologies and operational modes
Due to the constraints in transmission range, multi-hop self-organizing topologies
are essential in sensor networks, such as the IEEE 802.15.4. Devices are classified
with respect to their operation:
•
The Full Function Device (FFD) that contains a complete set of MAC services
and operates either as a simple network device or as PAN coordinator.
•
The Reduced Function Device (RFD) that contains reduced MAC services and
operates only as a network device.
The star topology is constructed around an FFD that acts as a PAN coordinator. The
FFD node is the only one that initiates links with more than one device. The peer-topeer topology is the second one allowed. Here, each device creates direct links to
other devices, thus generating redundant available paths.
An example of both the IEEE 802.15.4-compliant network topologies is displayed in
Figure 5.1 below.
61
City.Risks
Deliverable D2.1
Figure 5.1: The two IEEE 802.15.4-compliant network topologies: star and peer-topeer topology (Buratti et al, 2009).
The Star topology is preferred for covering a limited area and where small latency is
essential to the application. The communication is controlled by an FFD that acts as a
network master. Other network devices are allowed to communicate only with that
FFD. The predefined network policies establish which FFD can act as a network
master (PAN Coordinator) and form its own network.
In the case of covering large areas and when latency is not a critical issue, the peerto-peer topology is preferred. This topology forms more complex networks, where
any FFD can communicate with other FFDs inside its range via multi-hop. In this
topology, network devices are required to proactively search for neighboring
devices. When two devices are coupled they can exchange parameters to clarify the
type of services and features each one supports. A set-back for this topology is the
requirement for additional device memory for routing tables that is caused by the
multi-hop operation.
Other network topologies that are supported by IEEE 802.15.4 are cluster, mesh, and
tree. These last network topology options are described in the ZigBee Alliance
specifications (ZigBee Alliance, 2008).
All devices belonging to a particular network, regardless of the type of topology, use
their unique IEEE 64-bit addresses and a short 16-bit address is allocated by the PAN
coordinator to uniquely identify the network.
5.1.3. Mobile wireless sensor networks
Mobile Wireless Sensor Networks (MWSNs) are a new type of WSNs where the
sensors move. There are four possible entities types of entities in MWSNs, mobile
base stations, mobile relay nodes and mobile cluster heads (Shu et. al., 2012).
62
City.Risks
Deliverable D2.1
Moreover the movement of the sensors falls into three categories as well (Shu et. al.,
2012):
1. Controllable Movement, where the type of movement from the network
component is known and predetermined.
2. Predictable Movement, where the mobile sensor has a clear direction.
3. Unpredictable or Random Movement.
MWSNs usually display specific characteristics that include a dynamic topology,
increased energy requirements, unreliable communication links, and more accurate
localization. The mobility of sensors offers a number of advantages as well as
MWSNs display:
1. Long network lifetime, which is achieved via the dispersed transmission and
efficient energy consumption throughout the network. In classic WSNs the
nearest neighbor sensor of the gateway or sink usually is depleted faster.
That is not the case in MWSNs.
2. Increased Channel Capacity. Experimental studies suggest 3-5 times more
capacity gains than static WSNs, when the number of sinks increase linearly
with the number of sensors.
3. Enhanced Targeting since the sensors are not on static points but are
deployed randomly and are usually required to move for better sight or for
close proximity.
4. Enhanced Data Fidelity due to the limited number of hops.
A comparison of coverage methods is given in the table below.
Table 5.1: Coverage Methods Comparison (Zhu, Hara, Wang, 2014)
5.1.4. Coverage issues for MWSNs
Coverage is probably the most important parameter of MWSNs. It displays the total
area of network coverage, has a fundamental effect in the quality of service that the
network can provide and affects the application’s performance. Sensor coverage is
usually weakened due to disadvantageous initial deployments, sensor failures and
tough application environments and hardware limitations of sensors (e.g. limited
63
City.Risks
Deliverable D2.1
battery life). To tackle these disadvantages the sensors must have the ability to keep
coverage. This is achieved by two methods:
1. Self-deployment, where a sensor autonomously adapts its position to
improve coverage.
2. Relocation, where redundant sensors are moved to increase coverage.
5.1.5. Data management issues for MWSNs
Data collection is an essential task for any network. Different mobile devices have
different methods for collecting data. There are limited studies about data collection
using MWSNs, however three methods are the most prominent.
5.1.5.1. Mobile base station method
Using a Mobile Base Station with predictable mobility can result in significant power
savings (Chakrabarti, Sabharwal, Aazhang, 2003). However, the movement pattern
fo the base station is critical to maximize the efficiency. Linear programming
approaches (Wang et. al., 2005) have been adopted for finding an optimal pattern.
Other strategies conclude that the optimal movement follows the boundaries of the
network if the sensors are deployed in a circle (Luo, Hubaux, 2005). Finally, a mobile
base station with constraints in its route can perform data collection in a two-step
approach by firstly finding the minimum path in a discover phase (Gao, Zhang, Das,
2011)
5.1.5.2. Mobile relay nodes method
Mobile Relay Nodes Method utilizes relay nodes that store data from close range
sensors and act as buffers. The data are then dropped to access points or base
stations (Shah et al., 2003; Jain et al., 2006).
5.1.5.3. Mobile sensor nodes method
Two movement methods aim at improving the data collection (Shinjo et al. 2008):
1. In the Moving distance-based Static Topology the sensing node moves to a
predetermined position to join a gathering network and then communicate
with the base station.
2. In the Shortest Route with Negotiation utilizes a broadcast that informs the
sensing node about the position of other sensors that have already
connected to the base station.
64
City.Risks
Deliverable D2.1
City.Risks will come face to face with a number of problems in the area of Sensor
Communications.
Firstly, the protocols and frameworks under which the sensors will communicate will
be addressed in order to ensure the quality of service of the overall system. Since
the City.Risks sensor will be mobile the solutions adopted will address the issues of
coverage, for the «wake-up» signal of the sensor.
Secondly, the data management will also be addressed to enhance the possibility of
finding the stolen item even with limited data collected by other City.Risks devices
(Smartphones).
Those choices will be made in a manner consistent with the above methods.
5.2. Ground Monitoring and Theft Detection
5.2.1. Location finding
The generic portable communication device consists of a transceiver, an acceleration
sensor and a processor coupled to the acceleration sensor. The processor monitors
the acceleration profile of the portable device and compares it to at least one predetermined acceleration profile. Location finding includes steps that include the
monitoring of the device’s acceleration profile and entering a secure mode that
limits the access when a predetermined profile matches that of the device. The
method can also include a step where location is transmitted to an a-priori selected
target (e.g. voicemail, e-mail address, remote requestor with access code). The
location can be determined via GPS, time of arrival techniques or last known
location. The location data can also include a timestamp
An alternative method includes steps where a visual, audio or mechanical alert can
be sent to a cellular phone, hoping that the user will notice their misplaced phone
5.2.2. Existing monitor systems
WSN applications include structural health monitoring (bridges, buildings, etc.),
home security, intrusion detection, etc. The most common system architecture
installs stationary sensors at critical points and collects data that is transmitted to
the central computer for processing.
Figure 5.2 illustrates a WSN based real-time landslide monitoring system (Chen et.
al., 2008). This system was designed to predict dangerous geological phenomena
such as landslides, rock falls, and soil flow. The prediction is achieved by the constant
collection of data by different types of sensors such as inclinometers, tachymetry
and GPS. The data visualization is done through WebGIS, a web based geographical
information system. The data processing offers warnings through the web interface
and by using SMS if an indicative parameter gets below a specified threshold.
65
City.Risks
Deliverable D2.1
Figure 5.2: WSN based real-time landslide monitoring system (Chen et. al., 2008).
5.2.3. Collaborative target detection with decision fusion
Target Detection, tracking objects and surveillance are among the novel application
fields of WSN.
One of the main challenges of these applications is meeting the Quality of Service in
regards to increased detection probability, low number of wrong warnings and low
latency. To meet the demands it is often required to use large networks of sensors as
the conditions of application, and the spatiotemporal parameters are unpredictable
and dynamic. The loss of nodes due to sensor damage, or battery depletion is also
usual.
The WSN consists of static and mobile sensors as targets appear to pre-determined
«surveillance locations» following a certain probability. The network pinpoints the
«surveillance locations» after the deployment.
Nodes in the network self-organize into clusters around the surveillance spots by
running a clustering protocol (Chen, Hou, Cha, 2004), such that each cluster monitors
a surveillance spot. Mobile sensors can be part of multiple clusters, while a static
node belongs to only a single cluster.
The detection is implemented in two phases.
1. Phase 1: All sensors measure energy in sync. Local decisions are made by
comparing measurements to predetermined thresholds. The cluster head
receives reports from every sensor. A majority rule helps the cluster head to
make a system decision. Phase 2 is initiated after a positive decision.
2. Phase 2: Every mobile sensor moves towards the surveillance location. The
movement is consistent to a set of predetermined move list. The move
parameters are only the distance traveled and the instance the move begins.
Every sensor measures energy at a sampling interval. A sequential fusion-like
procedure is adopted. If the energy measured exceeds the predetermined
threshold at a specific sensor then a local positive decision is reported to the
cluster head. If the measurements are below the threshold, the mobile
sensor continues to move according its move-list and take measurements.
This described detection model offers the following benefits:
(1) Unnecessary movement of mobile sensors is avoided, as mobile sensors start
to move only after the first-phase detection produces a positive decision;
66
City.Risks
Deliverable D2.1
(2) The sequential fusion strategy allows each mobile sensor to locally control its
sensing and moving according to its movement schedule, which avoids internode coordination overhead. Moreover, a mobile sensor may terminate its
detection once it has enough evidence to make a positive decision. As a
result, the delay of reaching a consensus in the cluster can be reduced.
5.2.4. Public safety and surveillance networks in cities
Figure 5.3: Siklu’s EtherHaul-600T V-band 60 GHz radio and the EtherHaul-1200 Eband 70/80 GHz radio (Next Generation Wireless Security Application Note, 2015).
WSNs for the public safety and surveillance are expanding and provide more
applications. Moreover, the client-base for these applications is also increasing as
first-responders, law enforcement agencies and other public and private authorities
are rapidly adopting the enhanced capabilities of these services. The use of optic
fibers or leasing seems too costly. The same goes for microwave solutions due to
licensing. For these applications, the millimeter wave spectrum offers distinct
advantages in regards to the other costly solutions as well as the typically used sub-6
GHz, which lacks in throughput when compared with the Gigabit capabilities of
60/70/80 GHz bands. In urban areas it is also easier to deploy radios in the 60/70/80
GHz bands due to low interference, and in the 60Ghz unlicensed band is suitable for
deploying radios optimized for street level.
As a result, the typically used sub-6GHz band is limited and an expansion into
60/70/80 GHz will be needed to provide high-capacity and cost efficient alternative.
67
City.Risks
Deliverable D2.1
LTE technology for security applications will require additional networking resources.
LTE will increase the capabilities of first responders. This new capacity will be built on
a Gigabit backhaul network with carrier grade capabilities. Any future investment in
LTE security networks including FirstNet should address specifications for networking
and synchronization as implemented at carriers’ networks.
•
Fast and reliable deployment enabled by cascade and ring topologies.
•
Extended traffic monitoring and troubleshooting thanks to advanced
signalling (OAM).
•
High reliability and availability.
The City.Risks project is expected to tackle a number of challenges in the area of
target detection. These problems will address the frequency of the «wake up» signal
for the theft sensor as well as target detection from a limited set of data. The choices
of the consortium will be made in consistency with the above described methods.
5.3. Mobile Sensing in City.Risks
The project intends to develop a prototype identification sensor. City.Risks will
design and implement an innovative, small and discrete sensor coupling Bluetooth
Low Energy (BLE) and radio-based technologies to transparently identify and locate
stolen objects within a specific urban range through the usage of the City.Risks
network of citizens.
In particular, development process involves the following activities:
• Design and development of new BLE beacon device and enabling its
communication with a City.Risks mobile application.
• Integration of a Radio controlled Wake-Up mechanism to be used to
trigger a remote interrupt at a ‘sleeping’ beacon device, which can then
fire up its BLE to report its position to a nearby Mobile Application.
• Design and Development of Server Side Platform Components to
communicate with the BLE beacon devices and generate the Wake Up
Signals.
5.3.1. BLE relevant technologies
Apart from BLE, other technologies have been considered and assessed as core theft
detection technology, including the use of ZigBee and RFID. ZigBee is a more mature
wireless standard of the same class as BLE. Its Rx and Tx power are comparable to
BLE, and it also has very flexible topology in terms of ad hoc mesh networking
capabilities.
68
City.Risks
Deliverable D2.1
However, ZigBee suffers from high power consumption during idle listening and the
lack of direct compatibility with smartphones (smartphone or tablet). The only way
for ZigBee to reduce idle-listening cost is to duty cycle its Rx by software control, but
the average power is 10- 100 times that of BLE. To use ZigBee, the operator must use
either a custom-made ZigBee user-interface device to connect to the smart
containers, or use a smartphone to connect through a WiFi-ZigBee or BluetoothZigBee gateway node. The former is one additional device for the operator to carry,
while the latter would not work when the stolen item moves outside the range of
the gateway. Although a third option involving the use of a ZigBee dongle on the
smartphone is also possible, customers in general find dongles inconvenient, fragile,
and a burden.
Besides, wireless technologies may also be passive. Passive ones such as RFID allow
the tags to operate without batteries, since power is emitted by the RFID reader.
However, RFID readers can be relatively expensive and cannot support smarter
protocols in case of obstruction of RF signals by other containers.
Therefore it can be concluded that BLE has the necessary properties needed for our
application.
5.3.2. Bluetooth low energy technology overview
Wireless solutions are used in a variety of demanding industrial applications.
Technologies such as Wireless LAN, Classic Bluetooth, IEEE 802.15.4/ZigBee all provide
specific characteristics and are therefore suitable for different applications and specific
demands. However, none of these technologies offer an optimal solution for a wireless
connection for sensors and actuators in manufacturing automation. In these types of
applications, the existing technologies are too expensive, too slow or consume too much
energy. The solution lacks a fast, robust, low energy transmission for wireless sensors
and actuators. This is where Bluetooth low energy technology comes into play.
Bluetooth low energy (BLE) technology becomes particularly interesting as a wireless
technology option due to its ultra-low power consumption. Moreover it has the
advantage of being directly compatible with smartphones and tablets.
Bluetooth low energy is substantially different from Classic Bluetooth technology
which is ideal for continuous, streaming data applications including voice. Classic
Bluetooth has successfully eliminated wires in many consumer as well as industrial
and medical applications. Bluetooth low energy technology is ideal for applications
requiring episodic or periodic transfer of small amounts of data. Bluetooth low
energy has unique characteristics and new features that that are not practical with
Classic Bluetooth. For instance, coin cell battery-operated sensors and actuators can
now smoothly connect to Bluetooth low energy enabled smartphones, tablets or
gateways.
BLE represents one of the fastest-growing wireless technologies for the Internet of
Things (IoT). One reason for its popularity is its very long battery life: a slave can last
for one year on a CR2032 coin-cell battery while maintaining a logical connection
69
City.Risks
Deliverable D2.1
with a master. Second, it is directly compatible with smartphone devices
(smartphones and tablets) without requiring infrastructure or dongles. In the past
year, many BLE devices in the form of proximity tags, and low-power wearable
devices have been introduced to the market.
Power consumption is kept to a minimum as a Bluetooth low energy device is kept in
sleep mode most of the time and only wakes up when a connection is initiated. The
actual connection times are a few ms only, the maximum/peak power consumption
is less than 15 mA and the average power consumption is as low as 1 uA. It is
possible to power a small device with a coin cell battery – such as a CR2032 battery –
for several years. (Artem Dementyev 2013)
In order to achieve low power consumption, Bluetooth low energy uses a lower data
rate. In theory, this data rate is 1 Mbps but in practice the transfer rates for
Bluetooth low energy technology are less than 100 kbps. The following table
illustrates the basic features of classic Bluetooth and BLE.
Table 5.2. Classic Bluetooth and BLE basic characteristics
Data payload
throughput (net)
Robustness
Range
Local system density
Large scale network
Low latency
Connection set-up
speed
Power consumption
Cost
Classic Bluetooth
technology
2 Mbps
Bluetooth low energy
technology
~100 kbps
Strong
Up to 1000m
Strong
Weak
Strong
Weak
Strong
Up to 250m
Strong
Good
Strong
Strong
Good
Good
Very strong
Strong
For instance, Bluetooth low energy technology is new in having an efficient discovery
and connection setup, very short packets, asymmetrical design for small peripheral
devices and a client - server architecture.
5.3.3. BLE fundamental characteristics
This section summarizes the fundamental features of BLE technology.
5.3.3.1. Power consumption
Bluetooth low energy technology has been designed from the beginning to use the
lowest possible power consumption. For instance, the Bluetooth low energy unit can
be put in sleep mode where it is only used at an event of sending active files to a
70
City.Risks
Deliverable D2.1
gateway, PC or mobile phone. Further, the maximum/peak power consumption is set
to less than 15 mA and the average power consumption is at about 1 uA. A
foundation for the low energy consumption is the very fast connection set-up and
the short messages. Therefore, the energy consumption is reduced to a tenth of a
Classic Bluetooth unit.
5.3.3.2. Cost and backwards compatible
In order to be backwards compatible with Classic Bluetooth and to be able to offer
an affordable solution for very inexpensive devices, BLE chipsets are available in the
following two versions:
Dualmode:Bluetooth low energy technology as well as Classic Bluetooth
functionality.
Stand-alone: Bluetooth low energy technology only for optimizing cost,
power consumption and size, factors particularly useful for light discrete
battery powered devices.
5.3.3.3. Ease of use and integration
The technology uses a simple star topology, which simplifies the implementation
work significantly. This topology fits very well with common used system
architecture with a number of smaller devices connected to a master in a production
island. In most cases, an Infrastructure / Ethernet network is available and there is
no need for mesh networks to extend the geographical coverage. A unit is always
either a master or a slave, but never both.
The slave usually acts as an advertiser which keeps on advertising itself periodically
until a connection is established. The advertiser’s messages are generally destined
for a master that is listening to any advertising device in order to connect to it. The
communication between the master and the slave relies on the protocol stack, which
describes service group, roles and general behaviors. Services are collection of
characteristics and relationships to other services that encapsulate the behavior or
the device including hierarchy of services, characteristics and attributes used in the
attributes server.
5.3.3.4. Software structure
All parameters in Bluetooth low energy technology have a state that is accessed
using the so called Attribute Protocol. All attributes are represented as
characteristics that describe parameter value, presentation format, client
configuration, etc. Through these attributes, it is possible to build numerous basic
services and profiles. Some examples of basic services and profiles include the
following: Proximity, Automation I/O, Building Automation (Temperatures,
Thermostat, Humidity) Lighting (On/Off Switch, Dimmer), Remote Controllers,
Medical Devices (Pressure Meters, Scale Instruments etc.)
71
City.Risks
Deliverable D2.1
Connection and latency
Bluetooth low energy technology only uses three channels to build connections and
to discover other devices; this allows faster connection in only few msec and lower
power consumption.
With Bluetooth low energy technology, the latency periods are dependent on how
often the master sends messages to the slaves and how often it receives data from
the slaves. The latency period for one slave only is 7.5 ms and then increases slowly
for each additional slave.
5.3.3.5. Range
Thanks to the modulation scheme, Bluetooth low energy has an approximately 3 dB
better link budget compared to Classic Bluetooth. A Bluetooth low energy unit can
thereby offer a range of 100 meters in line of site without the need of an additional
power amplifier.
5.3.4. Beacons
5.3.4.1. What is a Beacon
Beacons are small, wireless hardware devices that transmit BLE signals via radio
waves to a personal device to enable a call-to-action which is customized for each
user.
Since Bluetooth’s launch, several companies have looked at the possibility of using
Bluetooth for advertising – pushing information to phones when the phone comes
within range of a fixed transmitter.
Although many companies exist within this area, it is an awkward experience that
leverages classic Bluetooth technology and it is poorly supported across phones. A
major drawback is that classic Bluetooth cannot broadcast messages to unpaired
phones – instead it needs to identify the presence of each individual phone and then
send a targeted message. To prevent multiple messages appearing, the transmitter
needs to keep a log of which phones it has previously sent messages to, incurring a
considerable level of complexity and cost.
BLE changes this by including a range of broadcast advertising modes. These are
fundamental to the technology and used for the discovery and pairing process,
essentially creating a much better user experience when pairing & connecting.
However, they can also be used for general, unacknowledged advertisements that
can be detected by any phone with its Bluetooth receiver turned on. It is this which
makes low cost Beacons possible.
The transmitted data is typically static but can also be dynamic and change over
time. With the use of Bluetooth low energy, beacons can be designed to run for
years on a single coin cell battery.
72
City.Risks
Deliverable D2.1
5.3.4.2. Device modes
A Bluetooth low energy device can operate in four different device roles. Depending
on the role, the devices behave differently. The first two roles are connection-based
a Peripheral device, which is assumed to be a low power device that exposes state or
information, and a Central device.
A Central is usually either a powered device, or one with significantly greater
processing capability and a rechargeable battery, e.g. a phone or tablet. Unlike
classic Bluetooth, the Peripheral and Central are very asymmetric in their resource
needs, with the standard being designed to minimize the complexity, power
requirements and costs of the Peripheral. In most cases, a Peripheral device spends
the majority of its life asleep, only waking when it needs to send data.
The other two device roles are used for one-directional communication:
• A Broadcaster is a non-connectable advertiser, for example, a temperature
sensor that broadcasts the current temperature, or an electronic tag for asset
tracking.
• An Observer scans for advertisements, but cannot initiate connections. This
could be a remote display that receives the temperature data and presents it, or
tracking the electronic tag.
The two obvious device roles for beacon applications are Peripheral and
Broadcaster. Both of them send the same type of advertisements with the exception
of one specific flag that indicates if it is connectable or non-connectable. A
Peripheral device that implements a GATT Server (GATT is an architecture for how
data is stored and exchanged between two or more devices) can be branded as a
Bluetooth Smart device. So a Bluetooth Smart branding indicates that the device is a
connectable Peripheral device that has data, which could be interacted with.
A Bluetooth low energy solution is ideal for beacons not only due to low power
consumption but also due to direct compatibility with smartphones while supporting
security features. The low-power consumption is accomplished by keeping the
transmission time short allowing the device to go into sleep mode between the
transmissions.
5.3.4.3. Non-connectable beacons
The non-connectable beacon is a Bluetooth low energy device in broadcasting mode.
It simply transmits information that is stored internally. Because the nonconnectable broadcasting does not activate any receiving capabilities, it achieves the
lowest possible power consumption by simply waking up, transmit data and going
back to sleep. This comes with the drawback of dynamic data being restricted to
what is only known to the device, or data being available through external input
73
City.Risks
Deliverable D2.1
from example serial protocols (universal asynchronous receiver/transmitter (UART),
serial peripheral interface (SPI), universal serial bus (USB), and so forth).
5.3.4.4. Connectable beacons
The connectable beacon is a Bluetooth low energy device in peripheral mode, which
means that it cannot only transmit, but also receive as well. This allows a central
device (for example, a smartphone) to connect and interact with services
implemented on the beacon device. Services provide one or more characteristics
that could be modified by a peer device. One example of these characteristic could
be a string of data that represents the broadcasted information.
5.3.5. Theft detection sensor technical consideration
From the previous description it is clear that BLE consumes much less energy than its
predecessors. Not only does it significantly extend the battery life of traditional
Bluetooth devices, but it also enables wireless communication for a class of lowpower devices that run on as little as coin-cell batteries. Besides, lower energy
consumption leads to a more robust, efficient product.
Because BLE is super energy-efficient, manufacturers have been aggressively building
BLE capability into modern devices such as phones, and tablets. On the supply chain
side, major manufacturers make BLE modules and chips widely available and in large
quantities. A wide and growing range of applications that communicate with BLEsupported devices are now available in both business and consumer markets.
However the most challenging issue as emerged from the project development
process is the Radio-Link mechanism (RLM) to be used for triggering a remote
interrupt at a ‘sleeping’ beacon device. The basic thing here is to combine BLE
module with an RF chipset operating in 2.4GHz or other frequency band.
Normally the radio network should be used to convey the Wake up signal from the
Web platform to theft detection sensor so as to wake up the BLE and then it should
broadcast the alarm signal. Of course radio network shall provide a local coverage
within a reasonable range of signal reception.
5.3.5.1. Sensor use case
The sensor shall be used in the following use case as described below:
A citizen attaches a theft detection sensor, a small and discrete sensor coupling
Bluetooth and radio-based technologies, to a personal item, e.g. mobile phone
or bicycle.
He / She registers the sensor with the authorities.
Once the item is stolen the citizen informs the authorities about the theft.
74
City.Risks
Deliverable D2.1
The responsible authority remotely activates the sensor from its hibernation
mode by multicasting a short-range signal that triggers the sensor to periodically
broadcast signals to mobile devices in proximity.
The signal broadcasted by the sensor is picked up by a mobile device with the
City.Risks mobile application installed and the authorities are notified by the
application that the stolen item has been located.
5.3.5.2. Operational scenario and working modes
BLE module should work in peripheral mode and in beacon mode. It shall simply
transmit information that is stored internally. Because the beacon device does not
activate any receiving capabilities, it achieves the lowest possible power
consumption by simply waking up, transmit data and going back to sleep.
Once an item (bicycle, motorbike, mobile asset) with such a module installed is
stolen, the owner shall inform the authorities which in turn shall send through an
application a broadcasting signal via Internet to certain local RF base station
infrastructure.
RF chipset –co-existing with BLE module on the same board-should be able to
receive a notification message from a nearby base station antenna within a certain
range of coverage.
Then through the interface between RF chipset and BLE the BLE module shall turn
from sleep mode to wake up mode and it will start transmitting the alerting signal to
the nearby mobile applications allowing for Detection of a stolen item in the area.
The amount of information that is going to be forwarded from RLM to BLE is not that
big, i.e. a wake up signal, therefore the interface should be quite simple and mostly
reliable. Radio chipset should normally work in receiver-only mode.
BLE should normally be in sleep/wake up mode as peripheral and should turn to
beacon mode only upon receiving alert message from the RF chipset. Again RF
chipset can also be put to sleep mode and then periodically wake up to be on
"listening-for-alert" mode. The last two working conditions shall be followed to
preserve module's battery.
5.3.5.3. Radio link mechanism approach
City.Risks shall define the required functions that the Radio-Link mechanism (RLM)
should encompass to trigger the ‘sleeping’ beacon device.
The radio link mechanism is an add-on that is included to the sensor in order to
formulate a compact, discrete battery-powered application.
Under this context, the development of such a prototype combining BLE module and
RF chipset must take into deep consideration that the integrated board should be
small-enough so as to be easily installed as a theft detection sensor.
75
City.Risks
Deliverable D2.1
RLM must be integrated in a chipset consuming as less energy as possible. Since the
power consumption is a key factor in sensor design, global RF networks such as GSM,
TETRA, UHF etc. although they might be suitable for functional prototyping they are
still too large as a deployment platform for our application, and are considered as
high energy -consuming devices.
A key point is the coverage distance that the RF network can cover. This certainly
means that if the user perceives the theft of his/her item by the time it has left the
coverage area then the broadcast signal originated by the authorities shall never
reach the module RF chipset as it will be out of coverage zone.
A fine-grained analysis is required so as to examine what could be the most suitable
radio link mechanism that can be coupled together with the BLE kit thus turning a
complete end-to-end powerful kit to powerful sophisticated theft detection sensor.
Under development process the most important aspects needed to be taken into
account are the following:
Integration of RLM with the BLE board
Definition of RLM and BLE inter-operational modes
Protocol handling.
The above aspects involve technologically-based subtasks. In particular the
development of the board is based on:
Schematics - How to wire things together
Layout - How to organize parts on a board
Manufacturing - How to get boards assembled in bulk
Code - How to establish communication from the BLE chip to RF chipset and
so forth.
The average power consumption of the RF chipset should also be kept very low
enabling quite a long time range operation. The power consumption is strongly
related to the operational logic and to the time periods where the radio mechanism
is “listening” for alarm broadcasting message and the BLE module is awake for
transmitting the alarm message to smartphones in proximity.
The use of Wi-Fi™ as an RLM is considered as a first choice option for our application
extending wireless connectivity to enable sensor’s communication with cloud
services and backend server, thus improving usability and reducing maintenance
needs.
Besides,2.4GHz unlicensed band and Wi-Fi™ are wireless standards with either a
specific focus on–or recent additions addressing–simpler but low power or ultra low
power wireless technologies. These are the main technologies that have been
examined as RF candidate mechanisms for the theft detection sensor
implementation.
The different technologies can roughly be split into the following categories:
Low power (average current consumption in a node 5-50+ mA):
• Wi-Fi™ direct, 2.4GHz chipsets
76
City.Risks
Deliverable D2.1
• Bluetooth® versions prior to v4.0.
Ultra low power (average current consumption <1 mA)
• ANT+™
• Bluetooth® v4.0 (which includes Bluetooth® low energy as a hallmark feature).
This segmentation also roughly corresponds to the applications that can use the
different standards. Due to the limited data transfer capacity, ultra low power
standards are mainly used in applications where the data throughput demand is low
(< 100 kbps) such as in sensor & actuator networks, user control input and for
limited size file transfers. This also applies to our specific application.
Lithium coin cell batteries, traditionally used for low power applications, are the
battery technology of choice for most of ultra low power wireless applications today.
These batteries are simple to fit in small enclosures and replacements are easily
accessible for the end user.
To this respect, as our application involves ultra low power wireless solution it is
absolutely imperative that the average current consumption shall be as low as
possible. But, focusing only on the average current assumes that the battery capacity
found in a battery data sheet is fixed for all conditions.
Even in sleep mode WiFi based RLM consumes much more power than BLE
technology resulting in sooner battery drainage. Besides once wake up, WiFi needs
to associate Access Point - this procedure might can take several seconds even to
minutes.
By examining all possible wireless candidate solutions that are available, it was clear
that critical point of sensor design involved the state when the radio circuitry is
activated. By that time it can draw large amount of current depending on
technology, vendor and implementation. This drain could far exceed the rated drain
current condition (~200 uA for a CR2032) for which the battery capacity is foreseen
to provide, according to the battery specifications.
5.3.5.4. Assessment of key parameters
There are specific key parameters which should carefully be taken into account:
Overall Size
The total size of the sensor must be kept very small since it is supposed to be placed
and attached to bikes, bicycles, hand-bags etc.
Total Power Consumption
Perhaps the most challenging issue here, since we must accommodate RLM kit along
with BLE kit, therefore a fine-grained analysis of power consumption is required.BLE
module should work as a peripheral device which is assumed to be a low power
device that exposes short piece of information. Despite of the peak current during
transmission can reach relatively high values the user should take into account that
these values are present only for small periods of time.
77
City.Risks
Deliverable D2.1
In most cases a peripheral device spends the majority of its life asleep only waking
up when it needs to send data. This is a good perspective though since the operation
of the BLE device is limited in only short time intervals.
Modular Scalable and easily customizable open platform
Application Logic structure should be scalable, expandable and allow for easy
configuration and over the air firmware download.
City.Risks theft detection sensor technical examination is currently in progress and
key findings have been already presented in previous paragraphs. In its current
phase, the on-going analysis of the state of the art includes the identification of
commercially available BLE hardware platforms that should be suitable for our
application development. The evaluation of characteristics, the consideration of
coupling mode between BLE and wake up radio mechanism, the design of
functionality and operational mode as well as the definition of application structural
logic have also been examined and addressed.
Planned future work include further hardware platforms performance
characterization, in-depth exploration of candidate wake up radio integration and
interoperability with BLE stack, and optimization of the implementation bottlenecks.
5.4. Existing Solutions and integrated projects related to City.Risks BLE
theft detection sensor
Under Task2.1 a survey of the state of the art has been performed in order to
pinpoint, analyze and evaluate existing projects, tools, applications, solutions and
case studies related to our project so as to identify gaps and challenges and address
new research directions.
In this section we summarize two existing complete and integrated solutions which
are related to City.Risks technical activities. The basic components and architecture
as well as a brief overview of each solution are described below pointing out the
fundamental features of each one.
5.4.1. BluVision/CycleLeash concept
With bicycle theft increasing at a rapid pace, CycleLeash (CycleLeash Case Study,
2014) developed a security system focused on bicycle anti-theft approach.
BluVision’s Bluetooth beacons are integrated into this solution. The innovative
sensor system communicates beacon reporting data whether a person has left the
desired area to a central cloud server. The solution utilizes BluFi, a Bluetooth to WiFi
gateway that are placed in several public locations enabling detection/alerting upon
a stolen bicycle in the area.
78
City.Risks
Deliverable D2.1
CycleLeash developed this project aiming at advanced bicycle tracking. The solution
incorporates end user smart phones and tablet devices, utilizing Bluetooth lowenergy sensors and mobile phone applications to provide bike tracking in real-time
environment. The advanced Bluetooth beacon sensor includes functionality that
allows cyclists to locate their bicycles in real-time through a smartphone application
or a cloud based web application. The application leverages information from
Bluetooth beacons, which detect the presence of bicycles.
CycleLeash utilizing BluVision’s beacons offers a reliable, cost effective solution for
accomplishing these activities. The CycleLeash is small, hidden and an effective
solution to combat the rise in thefts. It can safely be installed on a bike and with a
long lasting battery life, eliminating the need for recharging or replacing batteries.
Each CycleLeash has a beacon embedded into it with each Beacon having a unique
identifier that reports if the person has left the specified area. Through the
interactive Mobile Application the bicycle can be traced or help others locate theirs.
Figure 5.4: CycleLeash System Overview
The other component of the solution consists of a Bluetooth to WiFi Gateway. It can
transmit and receive data, provide remote OTA updates for entire deployments and
ensures that enterprise implementations are easy to manage by always collecting
available Bluetooth low energy data.
This Bluetooth low energy (BLE)-to-WiFi (BluFi) device eliminates the need for a
smartphone or tablet application to be actively running in proximity to scan and
discover beacons. The device is a part of a hub and can be managed to listen for any
Bluetooth beacon ID and report discovery within proximity ranges of the BluFi
sensor. One or more BluFi sensors listen for Bluetooth beacon data and connect to a
cloud-based server platform to receive notifications.
ID data enables the discovery of any enabled beacon within BluFi range. BluFi
delivers rapid development and seamless adoption with existing mobile beacon
applications and nearby WiFi routers.
79
City.Risks
Deliverable D2.1
BluFi features simple A/C plug-in to bridge any brand of beacon into existing WiFi
networks to receive and transmit Bluetooth beacons payload data over WiFi.
Software application provides set-up and receive beacon advertisements from BLE
beacons, transmitting them over WiFi to the backend software.
5.4.2. Bike Track concept
Due to their low cost and convenience, bicycles are a common means of
transportation in many countries. Bicycles have recently gained in popularity as an
environmentally friendly and healthy alternative to vehicles. Based on the increasing
number of bicycles, cyclists face the problem of not being able to find their bikes and
having their bikes stolen.
Scientific research has examined several technologies to cope with this critical
problem of stolen bicycles. RFID tags and GPS units are among the most important
projects for market exploitation. Those technologies have been widely used;
however, both of them involve high development cost or not adequate battery
autonomy.
To address this challenging approach, Dept of Computer Science and Information
Engineering, National Taiwan University presented BikeTrack (Tsung-Te Lai et al.,
2011) a participatory sensing system that uses everyday smartphones and low-cost
beacons to detect stolen bikes.
Each bicycle is equipped with a customized Bluetooth tag that actively sends
beacons. To discover the bicycle with a Bluetooth tag, participants use their mobile
phones to scan Bluetooth tags through an appropriate client application and report
the location of bicycle to a BackEnd Server application, so that bike owners can
locate their stolen bicycles.
5.4.3. BikeTrack fundamental principles
This section describes the basic design principles and choices:
Easy to deploy: BikeTrack project is challenging and easy to deploy. The
system is based on a smart phone application downloaded from a mobile app
download center (currently only at Android Market) and customizable
Bluetooth tags. Users act voluntarily for the scanning of bikes. The system
does not require any of authorities or agents involvement.
Accurate: the most important achievement is the accuracy of the reported
location when bike users cannot find their bikes. This project chose Bluetooth
technology because it offers a 10-20 meter range, which is sufficient to help
users locate their bikes. As for data reporting frequency, BikeTrack reports
periodically to ensure good coverage. This reporting frequency is based on
the mobility model of bike owners. For example, students often use their
bikes to commute between classes held in several large teaching buildings on
campus.
80
City.Risks
Deliverable D2.1
Minimal user overhead: To facilitate users to participate, BikeTrack project
has few advantages. For example, (1) the tags for bike owners do not require
care or battery recharge for 3–4 months, and (2) the client application has
minimal power consumption, CPU utilization, and network bandwidth to
avoid interfering with other.
5.4.4. BikeTrack platform components
Figure 5.5 illustrates an overview of BikeTrack system. The three main solution
components are (1) a bicycle equipped with a customized Bluetooth tag that
broadcasts a unique beacon ID, (2) mobile phones that run a BikeTrack client app to
scan the Bluetooth beacon from the bicycles, and (3) BikeTrack centralized server.
When a beacon is traced, the phone application logs the location, beacon ID, and
timestamp and sends the information back to the server. The BikeTrack server logs
user data to a database and provides a web interface that allows users to inquire
their bicycle locations on Google map. The following sections analytically describe
each component.
Figure 5.5: Bike Track system architecture
5.4.5. Customized Bluetooth tag installed on bicycle
Many types of radio channels, including RFID, Wi-Fi, Zigbee, or Bluetooth, can
potentially be deployed in bike tracking system. This study chooses Bluetooth
because (1) almost all mobile phones have a built-in Bluetooth technology, (2) the
81
City.Risks
Deliverable D2.1
radio range is up to 10-20 meters, and (3) the power consumption is low when it
operates only in discoverable mode.
Figure 5.6 (a) shows the actual class 2 Bluetooth device customized for this study by
a vendor. The main feature embedded in this Bluetooth device is that it runs only in
discoverable mode. Therefore no pairing is activated and only device name and MAC
address are continuously broadcasted. The Bluetooth device has a 40 to 50-day
lifetime on average. Figure 5.6 (b) shows how the device can be mounted on the
bicycle. The installation place has been chosen to be hidden to protect the device
from getting wet or being seen.
Figure 5.6: (a) Bluetooth tag, (b) mounted to a bicycle
5.4.6. Implementation of client app on mobile phone
The phone program in this study was designed on Android 2.x. The phone models
used in this study include HTC Hero, HTC Desire, HTC Legend, and Samsung Nexus 1.
When the user activates the application it scans nearby Bluetooth devices every 20
seconds. The client application only logs authorized Bluetooth tag by comparing the
MAC ID. If it finds a Bluetooth device, it logs the location (longitude and latitude),
timestamp, Bluetooth device name, MAC ID, and the user ID. Besides, phone battery
level and Bluetooth RSSI (receive signal strength indication) values are stored for
further optimization processing on battery consumption reduction and location
accuracy. The client application is designed to periodically upload relevant data to a
remote server. The battery lifetime is measured to be 22 hours for a client
application that scans for Bluetooth tags every 20 seconds on an HTC Desire HD.
5.4.7. Back-end software
The server runs an Apache HTTP server and MySQL database on a Linux machine.
This enhances data uploading and storage. A simple web interface is provided for
users to log in and pinpoint their stolen bicycle on Google Maps. Due to strict privacy
policy issues incorporated on the application, users are authorized to only view their
own assets.
The current tags can operate for 40-50 days. There has been a considerable
endeavor to extend tag lifetime so as to eliminate battery recharge during the
deployment period. One option is to increase the battery size from 800mA to
82
City.Risks
Deliverable D2.1
3000mA. This would result in lifetime extension from approximately 40 days to 150
days. Among other alternative options is Bluetooth duty cycle modification, or
energy induction from a small solar plane or even the motion of bike pedals.
Bluetooth Low Energy (BLE) technology is also a promising option, incorporating
many compelling features in terms of energy efficiency. The reason of its popularity
is its very long battery life : a slave can last for almost one year on a CR2032 coin cell
battery while maintaining a logical connection with a master, Second, it can provide
even longer radio transmission range (up to 50m to 100m) than the current
Bluetooth 2.0 standard. Therefore, BLE technology is a considered a more
appropriate replacement of the Bluetooth 2.0 that BikeTrack is currently
implementing.
83
City.Risks
Deliverable D2.1
6. Software Development Methodologies and Platform
Architectures
This section overviews the state-of-the-art in software development methodologies
and discusses system architectures relevant to software development tasks that
need to be completed in the City.Risks project.
6.1. Software Development Methodologies
The framework that plans and manages the development of an information system
is called a System Development Methodology. The specific technical needs of
projects have given birth over the years to a variety of frameworks, each with its
advantages and disadvantages.
Several software development Life Cycle Models have been formed and deployed to
provide this framework for planning and managing the development or modification
of a software product.
6.1.1. Design process models and theories
The most widespread design process models and theories are mentioned below.
6.1.1.1. Waterfall model
The “Waterfall Model” is a collection of Software Development Models (SDMs) that
breaks down the development of software into a sequence of phases. The sequence
includes:
1.
2.
3.
4.
The Analysis phase
The Design phase
The Coding phase
The Testing phase
Figure 6.1: Generalized Waterfall Model (Royce, 1970).
84
City.Risks
Deliverable D2.1
Variations of the “Waterfall Model” consist of different number of phases and
different interactions between each phase. Figure 5.1 displays three different
variations where solid arrows show forward-only version, solid and dashed lines
display variations that allow backtracking and solid, dashed and dotted lines indicate
a variation where the transition between any two phases is allowed.
The focus is on planning, scheduling, target dates and budgets for the
implementation of an entire system. Through the life-cycle of the project, formal
reviews and approvals/signoff by the beginning of the next phase create an
extensive written documentation.
6.1.1.2. The problem-design exploration model
The Problem-Design Exploration Model models the design as two interacting
evolutionary systems – the problem space P and the solution space S.
Figure 6.2: Problem-Design Exploration Model. P(t)=problem at time t; S(t)= situation
at a time t; dashed line indicates situation refocusing problem; diagonal downward
movement indicates a search process (Maher, Poon and Boulanger, 1995).
The advantages of Problem-Design Exploration Model are the simplicity and the
clear elucidation of the coevolution phenomenon that have been noticed in field
studies of designers (Cross, 1992).
This SDM was created to display the application of Genetic Algorithms to the
systems design, and does not show clearly how this is translated to human design
practice
6.1.1.3. Alexander’s design process models
The “Unselfconscious Process”, the “Selfconscious Process” and the “Formal
Process” were three SDM that were proposed by C. Alexander.
In the “Unselfconscious Process”, the designer eliminates oddities between form and
context by shaping the design object and other items directly.
85
City.Risks
Deliverable D2.1
In the “Formal Process”, the designer constructs a formal model using set theory and
exploits divide and conquer strategies to solve problems.
In the “Selfconscious Process”, the designer works by iterating between the
conceptual picture of the context and ideas and diagrams and drawings which stand
for forms”. This process fits the case where an independent agent pursues a goal by
taking actions in a self-determined sequence and monitors progress (Van de Ven and
Poole, 1995).
Figure 6.3: The Selfconscious Design Process. Arrows indicate interactions, not
sequence; Alexander does not specify the exact nature of these interactions
(Alexander, 1964).
6.1.1.4. Sensemaking-Coevolution-Implementation theory
In the Sensemaking-Coevolution-Implementation (SCI) SDM an agent develops a
complex software system with three basic processes, Sensemaking, Coevolution and
Implementation.
86
City.Risks
Deliverable D2.1
Figure 6.4: Sensemaking-Coevolution-Implementation Theory. Arrows indicate
relationships as shown, not a sequence of activities (Ralph, 2015).
Figure 6.5: Concepts and Relationships of SCI, Defined (Ralph, 2015)
6.1.1.5. Agile software development methodologies - an up-to-date approach
The increasing software complexity and dynamic user requirements have shifted the
focus of the software industry from the traditional SDMs to an agile based
development. The characteristics of agile methods are:
1. Shorter development cycles
2. Higher Customer Interaction
87
City.Risks
Deliverable D2.1
3. Incremental delivery
4. Frequent re-design necessitated by the changing in user requirements
This is unlike traditional software development approaches where there is focus in
extensive and thorough planning, process orientation, heavy documentation and
predictive approaches. This can be viewed in a number conducted of surveys
(Versionone, 2013).
The following table displays the differences between traditional methods and agile
methods.
Parameter
Traditional Methods
Agile Methods
Adaptability to Change
Change Sustainability
Change Adaptability
Development Approach
Predictive
Adaptive
Development Orientation
Process-Oriented
People- Oriented
Project Size
Large
Small/Medium
Planning Scale
Long-term
Short-term
Management Style
Command-and-control
Leadership-andcollaboration
Learning
Continuous Learning while Learning is secondary to
Development
Development
Documentation
High
Low
Agile based SDM result in better quality of software products as well as improved
productivity, flexibility, enhanced customer engagement and swiftness in changing
user requirements. Due to the above, the software development industry has rapidly
embraced agile approaches.
Extreme Programming, Scrum, Kanban, Lean, FDD (Feature-Driven Development),
Crystal, DSDM (Dynamic Systems Development Method), are some of the
methodologies that adopt the agile approach.
6.2. Platform Architectures for Emergency Management Systems
In modern advanced emergency management systems, many solutions have been
provided as attempts to support humans to take important decisions for very fast
recovery of critical situations through the consideration of various parameters and
events that occur concurrently. Critical situation detection is a very complex
procedure that involves both human and machine activities. The result of this
88
City.Risks
Deliverable D2.1
process is decision making for management and situation recovery purposes. Various
approaches have been studied in the literature. For instance (Itria et al., 2014)
studies a situation detection process that uses event correlation technologies
performing online analysis of real events through a Complex Event Processing
architecture. Such architectures are a suitable paradigm for the City.Risks platform,
since City.Risks will have to detect and analyze events that lead to decision support
procedures for further reactive measures. Event correlation is used to relate events
gathered from various sources, including crowdsensing and crowdsourcing, for
detecting patterns and situations of interest in the emergency management context.
The proliferation of modern mobile devices, such as smartphones and tablets, has
given a boost to the experimentation in the context of emergency management
systems allowing the integration of crowdsourcing and crowdsensing technologies
(Ganti et al. 2011). Crowdsourcing is the process of getting information online, from
a crowd of people, while crowdsensing refers to the involvement of a large, diffuse
group of participants in the task of retrieving reliable data from a specific field. By
means of the possibility to easily link persons, facts, events and places through a
large quantity of online geo-referenced data, users are the real holders of the “living
information” and the producers of current information about social phenomena and
dangerous events. This approach is also followed in the case of City.Risks where
users are considered as real “human sensors” providing qualitative and quantitative
information to the platform. The integration of information retrieved from mobile
devices, from social media and from several types of sensors deployed in the
infrastructure, allows the online analysis of a large amount of data used to detect
and identify dangerous events. Such online approach enables the detection of critical
situations as soon as they happen, so that a corresponding reaction can be
successfully performed. This mechanism aims to timely recognize critical situations,
usually called Real-time Situational Awareness (RTSA) (Beringer and Hancock, 1989).
The main goal of RTSA is to recognize the critical situations in the given application
domain as soon as possible in order to be able to take a decision for the necessary
measures to address them properly.
Decision-making is the first step of the reaction, and it should be made by humans
using a Decision Support System (DSS) that helps them decide how to face the
emergency. This process starts from data extraction and leads to the detection of
the situation in progress. Several challenges are introduced: (i) high efficiency, in
order to handle a huge amount of data and detect the situation as soon as possible
in order to allow time for a successful reaction (Cinque et al., 2013); (ii) it should be
able to detect critical situations before they happen (early warning) in order to
prepare a preventive action; (iii) it should be also tolerant to different types of noise,
meaning that the process should acknowledge only trusted information from trusted
sources, otherwise it could lead to wrong scenario definitions and consequently
wrong decisions; and (iv) it should be reliable to trust the logged events, including
architecture resilience and trustworthy data collection (Bondavalli et al., 2007;
Bondavalli et al., 2010), possibly allowing forensic analysis (Afzaal et al., 2012).
Complex Event Processing (CEP, (Beringer and Hancock, 1989)) technology aims to
resolve these challenges allowing an efficient management of the pattern detection
process in the huge and dynamic data streams and as such it is very suitable for
89
City.Risks
Deliverable D2.1
recognizing complex events and situations online. In (Beringer and Hancock, 1989),
the term event is defined as “an occurrence within a particular system or domain; it
is something that has happened, or is contemplated as having happened in that
domain”. This definition places the event concept into two different contexts: (i) the
real world in which events happens and (ii) the world of Information Technology
event processing, where the word event is used to mean a programming entity that
represents this occurrence. In the sphere of the Emergency Support Systems (ESS)
we consider events that happen in the real world and are represented in computing
systems through information entities. Event processing allows the detection of
critical situations in order to respond timely to the emergency. For the purpose of
our work, that is managing events constituted by textual description of facts and
involved entities (person or object), we define micro-events and complex-events.
Micro-events belong to a basic event taxonomy and represent simple real events
involving only one entity for example: people detection, fire presence, impulsive
sound recognition, object detection. Complex-events are the aggregation result of
the information contained in a set of micro-events, which are correlated by spatial,
temporal and causal relations defined by correlation rules. City.Risks will deploy
principles of the Complex Event Processing Theory and Practice in order to deploy
best practices for handling large data sets from various heterogeneous sources and
be able to detect early enough critical events for taking the appropriate measures.
Emergency response involves many people: rescue teams on the field, decisionmakers at different levels of government, citizens, press (Winter et al., 2005;
Zlatanova 2005). Their data needs vary accordingly but generally they are inline with
the big challenges of emergency response in urban context. Emergency response
should facilitate excellent coordination between different rescue groups/actors,
provide appropriate geo-referenced information and allow intelligence in
communicating orders and information to different participants. In general, we
consider three major types of system architecture that can be applied in such
emergency response systems. These are distinguished in: centralized (where all data
are processed and stored within the emergency center), federated (where the data
are maintained in a distributed fashion but under a common schema) and finally,
dynamic collaboration (meaning that data are stored in their original location and
format maintaining their original representation and relationships).
A system for an emergency center in urban areas requires among others i) the
discovery and effective use of various information sources (text, imagery, 2D and 3D
graphics, video) for monitoring and decision making, ii) the ability to provide
updated information to rescue units, decision makers and citizens fast, almost realtime, communication to transfer on-site information, iii) rapid determination of safe
evacuation routes considering dynamic factors such as current availability of exits,
stairs, etc. iv) the ability to alert communities at risk and provide information to the
general public, and others.
A system architecture for emergency response built on dynamic collaboration
principles is presented in (Zlatanova and Holweg, 2005). The system consists of three
levels: users, middleware and data on distributed servers. Some geospatial data
(such a topographic map, satellite images, cadastral maps and sewage network)
might be still maintained centrally, but most of the data will be distributed. Users are
90
City.Risks
Deliverable D2.1
grouped based on the type of equipment they would use to access information:
mobile, Virtual Reality centers and desktop users. All users access the system via a
given profile used to specify privileges, search for appropriate information and adapt
the delivered outputs. The major components of this set up are 'positioning and
communication middleware' (responsible for the contact with the users) and 'data
middleware' (responsible for information delivery). The two kinds of middleware
mentioned previously will be the connecting (integrating) components between the
different systems. City.Risks will offer as well a middleware layer upon which various
other components (for instance already existing software components in the
premises of the participating organizations) will be able to get integrated and
interoperated efficiently. However, its main principle of architectural design will rely
on a service oriented approach that will enable the communication with the actors
and the information delivery in a seamless way, agnostic of the underlying
technologies.
A number of modeling and simulation applications exist for studying individual
aspects of emergency response scenarios. There are solutions (Jain and McLean,
2004) that propose architectures for integrating geographically dispersed modeling
and simulation applications and repositories of data. The types of simulations
envisioned for emergency response as presented there, are multi-facetted, realtime, and synchronized, meaning that no single simulation model or software system
is capable of representing all aspects of the emergency response problem. Key
technical elements of the proposed solution are among others: Simulators, Scenario
Emulator, Data access and management components, Human interface modules and
Tools for identification, detection and learning. The interesting thing in this approach
is the positioning of the simulators which cover specific functionalities related to
simulating disaster events, impact, information flow etc. As far as it concerns the
data management framework a number of databases and libraries are used including
terrain and city street maps, land and building use databases, commuting pattern
databases, utility network databases, weather forecasts, etc. Real time data may be
generated by a number of automated sensors and from the inputs by various first
responders at the scene. Such data are also used for updating the simulators using
appropriate data processors (in their case XML) while they can be used for training
exercises and response to actual events. City.Risks architectural concepts have
similarities to the aforementioned approach. Simulator components will be deployed
in the Command Center that can simulate events that rarely occur (e.g. bombing or
terrorist attacks) familiarizing thus the personnel with managing the information
flows and the workflow orchestration for responding to the event effectively.
6.3. Research directions
The City.Risks software development task aims at developing the Operation Center
and its Core platform. In this context, one of the above methodologies will be
adopted to structure, plan, design and implement the software solutions of
City.Risks keeping in mind the user-centric vision of the consortium and fullfill its
goals for increasing the citizens' perception of security and safety in urban
environments.
91
City.Risks
Deliverable D2.1
7. Conclusions
We have surveyed the state-of-the-art in the research areas related to the City.Risks
project. Our analysis was conducted in two parts. First, complete and integrated
solutions that are relevant to City.Risks were described and analyzed. This analysis
overviewed a total of 22 projects and 46 software and hardware solutions. In the
second part, we focused on individual areas of research related to the project. In
particular we overviewed current research in the following areas:
Emergency Response and Risks Management
Data Management and Analytics
Mobile Sensors and Communications
Software Development Methodologies and Platform Architectures.
For each area of relevance, we outlined the most interesting current work with
respect to the goals of the City.Risks project, and proposed directions for future
work to be investigated within the scope of the project. More specifically, the results
of this analysis will be used to guide the project tasks responsible for the
implementation of the City.Risks platform and its components. These tasks include
WP3, Tasks 3.1-3.4, WP4, Tasks 4.1-4.4 and WP5, Task 5.1.
92
City.Risks
Deliverable D2.1
Appendix I. Rolled out applications
Name
URL
Short Description
Target Audience
Roles
Use case(s)
Urban Securipedia
http://securipedia.eu
Urban Securipedia is an urban security (and connected safety)
knowledge base which forms part of a complimentary tool with the
objective of assisting and supporting the urban planner to make
more informed deliberations and decisions on the proper planning
and sustainable development of the urban area, from a security (and
connected safety) perspective. Urban Securipedia assists the urban
planner in making more informed decisions in the first (concept
level) of three distinguished levels of urban planning. It is part of a
complete tool set to support each of these levels, including concept
level tools, plan level tools and detail level tools.
Target Audience
Urban planners
Role
Information
Features
Search by subjec, search by
urban object type, Text free
search, knowhow
PEP
http://crisiscommunication.fi/pep
The project Public Empowerment Policies for Crisis management
(PEP, 2012–2014) identifies best practices in a community approach
to crisis resilience, and gives directions for future research and
implementation, including the use of social media and mobile
services. The input of experts in the field of crisis management and
communication is a key element in pursuing the goals of this project.
Target Audience
policy makers, Decision makers,
researchers, and
crisis management
and communication
experts
Role
Information
Features
Score card audit, Search by
subject
COMPOSITE
http://www.composite-project.eu
COMPOSITE – short for “Comparative Police Studies in the EU” – is a
research project that looks into large scale change processes in
police forces all over Europe. New types of crime, open borders, new
technologies, changing public expectations and tighter financial
resources are directly or indirectly affecting police forces in most
European countries. These new demands require modern police
forces that are managed efficiently, are capable of acting flexibly and
have the means to cooperate with forces in other countries. Many
police forces respond by introducing ambitious change programmes,
aiming at modernising and rationalising the way policing is
conducted. As such the face of European policing is slowly changing.
Target Audience
Policy Makers
Role
Information
Features
Not found
PACT
http://www.projectpact.eu/
Public perception of security and privacy: Assessing knowledge,
Target Audience
general public
Role
93
City.Risks
Deliverable D2.1
Collecting evidence, Translating research into action
Information
Features
Assessment of Risk
Alert4All
http://cordis.europa.eu/project/rcn/98427_en.html
Alert4All focuses on improving the effectiveness of one element of
the People-Centred Early Warning Systems paradigm, namely alert
and communication towards the population in crises management.
This improvement shall be measurable in terms of cost-benefit ratio,
number of affected citizens timely reached by alerts, trust of citizens
on alerts and intended vs. actual impact of alert strategies.
Target Audience
Public Protection, general
public
Role
Communication
Features
Information Management
Portal, New Media Screening,
Alert Simulation,
Communication System
SUBITO
http://cordis.europa.eu/project/rcn/89391_en.html
The SUBITO programme has been developed to address Theme
Detection of Unattended Goods and of Owner. It will focus on the
automated real time detection of abandoned luggage or goods and
the fast identification of the individual who left them and their
subsequent path.
Target Audience
Public Place Control Center
Role
Communication
Features
Alarm System, Person Tracking
INDECT
http://www.indect-project.eu/
The purpose of the INDECT project is to involve European scientists
and researchers in the development of solutions to and tools for
automatic threat detection.
The primary objective is to develop advanced and innovative
algorithms for human decision support in combating terrorism and
other criminal activities, such as human trafficking, child
pornography, detection of dangerous situations (e.g. robberies) and
the use of dangerous objects (e.g. knives or guns) in public spaces.
Efficient tools for dealing with such situations are crucial to ensuring
the safety of citizens.
Target Audience
Police Forces
Role
Information
Features
Thread Detection Monitor,
Computer Network Thread
Detection. Data and Privacy
Thread Detection
SECUR-ED
http://www.secur-ed.eu/
The SECUR-ED Project was a demonstration project with an objective
to provide a set of tools to improve urban transport security.
Participants included all the major stakeholders from across Europe.
Based on best practices, SECUR-ED integrated a consistent,
interoperable mix of technologies and processes, covering all
aspects; from risk assessment to complete training packages. These
solutions also reflected the very diverse environment of mass
transportation and also considered societal and legacy concerns.
Target Audience
Passengers, Front-line and
Security employees, Operators
in control centres, Security
managers, decision makers
Role
Information, communication
Features
packaged modular solutions for
mass transport security
THALES Integrated and scalable urban security solutions
https://www.thalesgroup.com/en/worldwide/security/integratedand-scalable-urban-security-solutions
Thales solutions include conventional architectures incorporating
legacy assets, secure cloud architectures relying on the latest
Target Audience
City Authorities and Agencies
Role
Information, Communication,
Collaboration
94
City.Risks
Deliverable D2.1
virtualisation technologies, and platform maintenance and cyber
protection to support customer-operated services. Many of Thales’s
operational services are also now available through new delivery
models (e.g. video surveillance as a service)
Features
Emergency Management,
Dispatch and management of
security forces, Mobile
Command & Control, Apps for
intervention forces, Video
surveillance and sensor
systems, Video analytics,
Citizen applications, Leverage
Big Data
SAMSUNG Urban Security Systems
http://www.samsungsecurity.com/_img/menu4/Urban_security_Sol
ution_bro_eng_0326F.pdf
As cities grow and become more complex, accidents and various
threats are on the rise. To protect human life and facilities from
these threats and to create a more secure and comfortable
environment, Samsung Techwin provides cutting-edge security
solutions.The TSM integrated platform can analyze numerous types
of data to help make a quick decision and immediate response, to
any incident in and outside the city this contributes to enhanced
security for the entire city while reducing costs.
Target Audience
Public Security
Role
Information
Features
Data Collection, Control Center,
Response Measures
SAAB SAFE Emergency Response
http://saab.com/security/land-transport-and-urban-security/urbansecurity-solutions/safe-security-management-system/
Saab Security and Safety Management is offering you a solution that
will improve workflows and create more efficient processes while
increasing security and safety. It will provide employees with a brand
new user experience and deliver the powerful resource management
your operations need. We have identified current and future
demands, and we are taking care of them for you, today.The core of
the Saab Security and Safety Management Solution is SAFE, a flexible
platform for building next generation security systems. One system
managing all areas of your operations
Target Audience
Missioncritical operations
Role
Information
Features
Workflow management,
Infrastructures, Operator client,
Maps, Resource Management,
Sensors, Video,
Communication, Mobile,
Business Intelligence,
Integration
Selex ES Urban Security Video Surveillance
http://www.selexelsag.com/internet/localization/IPC/media/docs/Vi
deosorveglianza_SelexES.pdf
The security of sensitive areas (government offices, industrial sites,
military zones, prisons, stadiums, banks, critical urban areas, ports,
airports, railway stations, metro stations) is vital to safeguard
personal safety and protect environments, systems, information and
data inside these areas. Recent history teaches us that electronic
protection systems aren’t only a precautionary measure, but an
effective tool for preventing and deterring criminal acts.
Target Audience
Public Security
Role
Information
Features
Surveillance
Selex ES CITIESvisor
http://www.selexelsag.com/internet/localization/IPC/media/docs/CI
TIESvisor.pdf
Public transport companies are increasingly warning of the need to
improve safety on their vehicles to combat vandalism and crimes
Target Audience
Public Transport Organizations
Role
Information
Features
95
City.Risks
Deliverable D2.1
against passengers and drivers. An effective way of preventing this
kind of behaviour is using on-board television cameras which are
dual purpose: a deterrent for the criminal and an investigative tool
for any inquiries
Surveillance
TAS-AGT Urban Security
http://www.tas-agt.com/urban-security/
Our Urban Security platform supports multi-agency collaboration
and information sharing, generating a comprehensive Unified
Situation Awareness Picture (USAP) that empowers law enforcement
and counter-terrorism officials to take effective and timely action. By
utilizing visual sensors equipped with advanced video analytics,
dedicated CBRNE (chemical, biological, radiological, nuclear,
explosives) detectors, as well as data from participating citizens who
deliver real-time information via tools such as smart phones, our
solution enables authorities to provide city residents with a safe and
secure urban environment.
Target Audience
City Authorities and Agencies
Role
Collaboration
Features
Surveillance, Intelligence
System, Citizen Data Collection,
Visualisation, Investigation,
Threat Identification
LL Tech International Urban Security
http://www.lltechinternational.com/our-solutions/urban-security/
The Safe City is a concept for returning security, safety and quality of
life to today’s complex cities through the use of technology,
infrastructure, personnel and processes. The Safe City concept can
be applied to cities, towns, industrial parks, college campuses, or any
other physical environment where people require a safe,
comfortable environment.
Target Audience
municipal and national
decision-makers
Role
Information
Features
Detection of irregular activities,
response and reaction
measures
ISS SecurOS
http://isscctv.com/
The ISS SecurOS solution set powers the most advanced video
management and video analytics deployed anywhere in the world.
From the warning and aversion of threats, to the prevention of
terrorism and the provision for safety of people and economic wellbeing of businesses. ISS is the proven technology partner of the
world’s largest integrators in the video security and surveillance
marketplace. ISS has a very wide range of deployments, in areas such
as transportation, retail, banking, colleges, government, industry,
and urban surveillance, and with a true open platform that allows
one to expand their solutions at their own pace.
Target Audience
Public Security
Role
Information
Features
Video Management System,
License Plate Recognition
(LPR/ANPR), Container
Recognition, Face Capture and
Recognition, Carriage
Recognition for Rail, Traffic
Monitoring
ARMOR
http://create.usc.edu/sites/default/files/projects/sow/797/tambeor
donez2008-assistantforrandomizedmonitoringoverroutesarmor.pdf
The ARMOR project is focused on developing methods for creating
randomized plans and processes for monitoring, inspection,
patrolling, and security in general – so that even if an attacker
observes the plans, he/she cannot predict its progression – thus
providing risk reduction while guaranteeing a certain level of
protection quality.
Target Audience
LAWA Police
Role
Information
Features
model patrols of an agent/team
of agents
Emexis Fuel Tracker
Target Audience
96
City.Risks
Deliverable D2.1
http://www.emixis.com/technology-for-telematics-serviceproviders/fuel-tracker-fuel-theft-detection-and-fuel-levelmeasurement/
EMIXIS’ Fuel Tracker module allows measurement, both remotely
and in real time, fuel levels, as well as detection of theft by
siphoning. This very simple to install universal module can be
mounted on all types of vehicles, it requires no special calibration or
connection other than to the vehicle tank gauge. Its theft detection
function is based on a unique self-learning mechanism and allows
local action (activation of the horn or headlights, for example). In
addition, it can be connected to any market GPS beacon which
features an analog input in order to transmit the level of fuel of the
vehicle.
Businesses
Role
Information
Features
Fuel Tracking, Position Tracking
CrimeReports
https://www.crimereports.com/
Crime Map and Visualizaiton
Target Audience
general public
Role
Information
Features
Location Selection
UKCrimeStats
http://ukcrimestats.com
Crime Data Repository
Target Audience
general public
Role
Information
Features
Different Data Filters
Crime Map Vienna - Kriminalität in Wien
http://www.vienna.at/features/crime-map
Target Audience
general public
Role
Information
Features
Location Selection, Crime Type
Selcetion
Durham Crime map
https://www.police.uk/durham/39/crime/
Target Audience
general public
Role
Information
Features
Location Selection
VirtualGuard
http://www.virtualguard.com/
Virtual guard security company can offer you reliable security
solutions that will protect your premises and give you peace of mind.
With our advanced video surveillance technology and our world class
Control and Command system, we can provide you with security
services that are faster and more reliable than those of other
security companies. Our security services can help you to: Prevent
theft and vandalism, Prevent crime before it happens, Prevent
Target Audience
general public
Role
Information
Features
Video Surveillance, Audio
Warning, Command & Control
Center
97
City.Risks
Deliverable D2.1
intruders from entering the premises & Monitor everything that
goes on in the premises at all times
BluCop HappstoR
http://blucop.happstor.com
BluCop uses the communication cost free Bluetooth technology to
periodically check the availability of previously selected Bluetooth
devices - so called markers - in the surrounding area. BluCop raises
the alarm if it does not receive a life sign response of the currently
checked marker within a configurable time. As Bluetooth is a short
range communication technology, the alert is issued whenever the
checked marker gets too far from BluCop.
Target Audience
general public
Role
Information
Features
Device Monitoring
Prey
https://preyproject.com/
Prey is an all-in-one cross-platform security solution for laptops,
tablets, and phones used to protect millions of devices, all around
the world. Its solid tracking and reporting technology helps people
and organizations keep track of their assets 24/7, and proves to be
crucial in the recovery of stolen devices every day
Target Audience
general public
Role
Information
Features
Tracking, Locking, Memory
Erasure
Comodo Anti Theft
https://play.google.com/store/apps/details?id=com.comodo.mobile.
comodoantitheft
Where is my Phone? – Don’t worry about your phone even if it’s lost
or stolen. You can control your devices remotely from any place, any
time.
Target Audience
general public
Role
Information
Features
Activity Log, Uninstall
Protection, Localization,
Remote Lock, Alarm, Memory
Erasure, Remote Take a Picture
98
City.Risks
Deliverable D2.1
Appendix II. Ongoing projects’ development
Name
URL
Short description
Target Audience
Roles
Use case(s)
ATHENA
http://www.projectathena.eu/
Athena is a system that harnesses social media and high-tech mobile
devices to crowd-source information during a crisis. It combines this
information with the domain knowledge of the emergency services
and novel analytical techniques to provide the public and first
responders actionable intelligence and map-based visualisations to
help safeguard and rescue citizens caught up in crisis situations.
Information from the ‘crowd’ is processed, enhanced and retuned
back to the crowd – ‘collective intelligence’ that can transform
disorganised individuals into organised pre-first responders.
Target Audience
first responders, citizens
Role
Collaboration
Use cases
Crisis management, Social media
in crisis management
eVACUATE
http://www.evacuate.eu
A holistic, scenario-independent, situation-awareness and guidance
system for sustaining the Active Evacuation Route for large crowds.
Target Audience
general public
Role
Information, Communication
Use cases
Safety/Security
TACTICS
http://fp7-tactics.eu/
Tactical Approach to Counter Terrorists in Cities
Target Audience:
Threat Manager (TM), Threat
Decomposition Manager (TDM)
and Capabilities Manager (CM)
Role
Information
Use cases
Legal, Safety/Security
HARMONISE
http://harmonise.eu/
A Holistic Approach to Resilience and SysteMatic ActiOns to Make
Large Scale UrbaN Built Infrastructure Secure
Target Audience:
Mechanisms/Tools for Delivery
of Improved Urban Security and
Resilience
Role
Information
Use cases
Safety/Security
Urban Security eGuide (Inspirational Plattform)
http://www.besecureproject.eu/dynamics//modules/SFIL0100/view.php?fil_Id=56
The main objective of the Inspirational Platform (IP) is to provide a
capability to browse efficiently BESECURE data resources (case study
best practices, urban security related literature reviews) to gain
additional knowledge.
Target Audience:
Education, community, general
public
Role
Information
Use cases
Safety/Security
99
City.Risks
Deliverable D2.1
Policy platform
One of the main objectives for the Policy Platform is to provide a
capability to create so called “One Page Policies (OPP)” being an
executive summary of the emerging evidence-based urban security
policies. Such OPPs can together with attached more detailed
evidences be used as a proposals for hanges for high-level
authorities with a possibility to have an insight to the whole policy
development process whenever needed.
Target Audience:
Policy makers
Role
Collaboration
Use cases
Urban security Early warning system
The purpose of D4.3 is to provide a prototype of an Early Warning
System that provides an overview of the security situation in an
urban area, and that supports the monitoring of various factors
characterising urban zones. The system can provide alerts of certain
factors change, so that policy makers can carry out timely
countermeasures against undesirable scenarios.
Target Audience:
Policy makers
Role
Information
Use cases
iRISK Urban Vulnerability Measure
http://create.usc.edu/sites/default/files/projects/sow/1045/kurbanc
reateyear8annualreportkurbanhudoc.pdf
Prototype web-based application of Personal Vulnerability Index for
pilot counties in North Carolina is currently under development. The
PI has completed the PVI module for IHRM Loss Estimation project in
Raleigh North Carolina. The intended users are general public and
local and state stake holders. The user will enter information on
household characteristics, housing type and disaster strength and
estimate uncovered structure losses both in terms of dollars and
lossrates.
Target Audience:
NC state policy makers, Howard
University students
Role
Information
Use cases
Economic, Social,
Safety/Security
RAW Risk Assessment Workbench
http://create.usc.edu/sites/default/files/projects/sow/850/hall2005riskanalysisworkbenchpart1.pdf
The Risk Analyst Workbench (RAW) is a software tool that provides
modeling and analysis capabilities for the risk analysis and decision
analysis steps of CTMS (CREATE Terrorism Modeling System). RAW
also provides a mechanism for extracting data from external sources,
building libraries of data for internal use and linking models to
support other modeling steps. RAW guides the risk analyst through
the steps of threat and counter-measure characterization,
probability estimation, outcome definition, and scenario creation. It
also provides tools for rating outcomes of threats, effectiveness of
counter-measures, and prioritizing investments.
Target Audience:
classified, “official use only” or
public environment
Role
Information
Use cases
Safety/Security
DPS Deploy
Target Audience:
http://create.usc.edu/researcher/michael-orosz/projects/dpsdeploy- USC Department of Public Safety
usc-department-public-safety-risk-assessment-and
Role
This project will develop risk-based crime prediction and
Information
countermeasure allocation models and tools to be used by the USC Use cases
100
City.Risks
Deliverable D2.1
DPS to facilitate decision making process. In addition, the team will
focus on generalizing the technology for use in other physical
infrastructure environments such as maritime port (i.e., update
PortSec analytics), stadium and other physical infrastructure
operations.
Safety/Security
MobEyes
http://lia.deis.unibo.it/Research/Mobeyes/
MobEyes exploits wireless-enabled vehicles equipped with video
cameras and a variety of sensors to perform event sensing,
processing/filtering of sensed data, and ad hoc message routing to
other vehicles. Since the sheer amount of data will be generated
from those sensors, directly reporting raw data to the authority is
infeasible. Thus, MobEyes proposes that: sensed data stay with
monitoring mobile nodes (i.e., mobile storage); vehicle-local
processing capabilities are used to extract features of interest, e.g.,
license plates from traffic monitoring images; mobile nodes
periodically generate data summaries with extracted features and
context information such as timestamps and positioning
coordinates; mobile agents, such as police patrolling cars, move and
opportunistically harvest summaries from neighbor vehicles. The
harvesting agents are interested in the following data: where was a
certain vehicle at a certain time; which vehicles were at a given time
in a given place, and: what data/video did the vehicle(s) collect? To
access the data later, one needs to get to the actual vehicles (based
on summary reports) and pump out the data.
Target Audience:
urban monitoring
Role
Information
Use cases
Safety/Security, Mobility
101
City.Risks
Deliverable D2.1
Bibliography
Abraham, I., Delling, D., Goldberg, A. V., Werneck, R. F. (2011). A hub-based labeling
algorithm for shortest paths in road networks. In Experimental Algorithms (pp. 230241). Springer Berlin Heidelberg.
Abul, O., Bonchi, F., Nanni, M. (2008). Never walk alone: Uncertainty for anonymity
in moving objects databases. In Data Engineering, 2008. ICDE 2008. IEEE 24th
International Conference on (pp. 376-385).
Afzaal, M., Sarno, C. D., Coppolino, L., D'Antonio, S., Romano, L. (2012). A resilient
architecture for forensic storage of events in critical infrastructures. In HighAssurance Systems Engineering (HASE), 2012 IEEE 14th International Symposium on
(pp. 48-55). IEEE.
Agarwal, P., Vaithiyanathan, R., Sharma, S., Shroff, G. (2012). Catching the Long-Tail:
Extracting Local News Events from Twitter. In ICWSM.
Agatz, N., Erera, A., Savelsbergh, M., Wang, X. (2012). Optimization for dynamic
ride-sharing: A review. European Journal of Operational Research, 223(2), 295-303.
Akiba, T., Iwata, Y., Yoshida, Y. (2013). Fast exact shortest-path distance queries on
large networks by pruned landmark labeling. In Proceedings of the 2013 ACM
SIGMOD International Conference on Management of Data (pp. 349-360). ACM.
Alexander, C.W. (1964). Notes on the Synthesis of Form. Harvard University Press.
Aliance, Z. Zigbee Specifications (2008). Zigbee Standard Organisation, San Ramon,
CA, USA, 2008.
Alivand, M., Hochmair, H. (2013). Extracting scenic routes from VGI data sources. In
Proceedings of the Second ACM SIGSPATIAL International Workshop on
Crowdsourced and Volunteered Geographic Information (pp. 23-30). ACM.
Anderson, D.J. (2010). Kanban: Successful Evolutionary Change for Your Technology
Business, Blue Hole Press, Ed.1.
Bae, S., Kim, D., Ha, M., and Kim, S. H. (2011). “Browsing architecture with
presentation metadata for the internet of things,” in Proceedings of the 17th IEEE
International Conference on Parallel and Distributed Systems (ICPADS 11), pp. 721–
728.
Banks, B., Harms, T., Sarvestani, S.S., Bastianini, F. (2009) A low-cost wireless system
for autonomous generation of road safety alerts, in: SPIE Smart Structures and
Materials+ Nondestructive Evaluation and Health Monitoring. International Society
for Optics and Photonics, p. 72920E–72920E.
Barsky, L. E., Trainor, J. E., Torres, M. R. and Aguirre, Benigno. E. (2007). Managing
volunteers: FEMA's Urban Search and Rescue programme and interactions with
unaffiliated responders in disaster response. , Disasters, 31: 495–507.
102
City.Risks
Deliverable D2.1
–Hannemann, M., Pajor, T., Sanders, P.,
Wagner, D.,
Werneck, R. (2014) Route planning in transportation networks.
Technical Report MSR-TR-2014- 4, Microsoft Research.
Bast, H., Funke, S., Sanders, P., Schultes, D. (2007). Fast routing in road networks
with transit nodes. Science, 316(5824), 566-566.
Beck, K., et al. (2001).,
http://Agilemanifesto.org/
Manifesto
for
Agile
Software
Development.
Becker, H., Iter, D., Naaman, M., Gravano, L. (2012). Identifying content for planned
events across social media sites. In Proceedings of the fifth ACM international
conference on Web search and data mining (pp. 533-542). ACM.
Becker, H., Naaman, M., Gravano, L. (2011). Beyond Trending Topics: Real-World
Event Identification on Twitter. ICWSM, 11, 438-441.
Belimpasakis, P., You, Y., Selonen, P. (2010). Enabling rapid creation of content for
consumption in mobile augmented reality. In: Next Generation Mobile Applications,
Services and Technologies, 2010 Fourth International Conference on, IEEE, 1-6.
Beresford, A. R., Stajano, F. (2003). Location privacy in pervasive computing. IEEE
Pervasive computing, (1), 46-55.
Beringer, D., Hancock, P. (1989). Exploring situational awareness: A review and the
effects of stress on rectilinear normalization((aircraft pilot performance)). In
International Symposium on Aviation Psychology, OH (pp. 646-651).
BluVision (2015). Bluetooth to WiFi Gateway, http://bluvision.com/blufi-wifi-sensor
Bollen, J., Mao, H., Zeng, X. (2011). Twitter mood predicts the stock market. Journal
of Computational Science, 2(1), 1-8.
Bondavalli, A., Ceccarelli, A., Falai, L., Vadursi, M. (2007). Foundations of
measurement theory applied to the evaluation of dependability attributes. In
Dependable Systems and Networks, 2007. DSN'07. 37th Annual IEEE/IFIP
International Conference on (pp. 522-533). IEEE.
Bondavalli, A., Ceccarelli, A., Falai, L., Vadursi, M. (2010). A new approach and a
related tool for dependability measurements on distributed systems.
Instrumentation and Measurement, IEEE Transactions on, 59(4), 820-831.
Bontcheva, K., Rout, D. (2012). Making sense of social media streams through
semantics: a survey. Semantic Web, 1, 1-31.
Boulos, M.N.K., Resch, B., Crowley, D.N., Breslin, J.G., Sohn, G., Burtner, R., Pike,
W.A., Jezierski, E., Chuang, K.-Y.S., 2011. Crowdsourcing, citizen sensing and sensor
web technologies for public and environmental health surveillance and crisis
management: trends, OGC standards and application examples. International journal
of health geographics 10, 67.
Buratti, C., Conti, A., Dardari, D., Verdone, R. (2009). An overview on wireless sensor
networks technology and evolution. Sensors, 9(9), 6869-6896.
103
City.Risks
Deliverable D2.1
Cameron, M. A., Power, R., Robinson, B., Yin, J. (2012). Emergency situation
awareness from twitter for crisis management. In Proceedings of the 21st
international conference companion on World Wide Web (pp. 695-698). ACM.
Cao, L., Jiang, W. and Zhang, Z. (2009). Automatic Meter Reading System Based on
Wireless Mesh Networks and SOPC Technology’, Intelligent Networks and Intelligent
Systems.
Cao, X., Cong, G., Jensen, C. S. (2010). Retrieving top-k prestige-based relevant
spatial web objects. Proceedings of the VLDB Endowment, 3(1-2), 373-384.
Cao, X., Cong, G., Jensen, C. S., Ooi, B. C. (2011). Collective spatial keyword querying.
In Proceedings of the 2011 ACM SIGMOD International Conference on Management
of data (pp. 373-384). ACM.
Cao, X., Cong, G., Jensen, C. S., Yiu, M. L. (2014). Retrieving regions of interest for
user exploration. Proceedings of the VLDB Endowment, 7(9), 733-744.
Cardei, M., Thai, M. T., Li, Y., Wu, W. (2005, March). Energy-efficient target coverage
in wireless sensor networks. In INFOCOM 2005. 24th Annual Joint Conference of the
IEEE Computer and Communications Societies. Proceedings IEEE (Vol. 3, pp. 19761984). IEEE.
Cha, K. (2001). Wireless remote monitoring system for reservoir management. Proc.
of Industry-University Collaboration, 4, 99-106.
Chainey, S., Ratcliffe, J. (2013). GIS and crime mapping. John Wiley Sons.
Chainey, S., Tompson, L., Uhlig, S. (2008). The utility of hotspot mapping for
predicting spatial patterns of crime. Security Journal, 21(1), 4-28.
Chakka, V. P., Everspaugh, A. C., Patel, J. M. (2003). Indexing large trajectory data
sets with SETI. Ann Arbor, 1001, 48109-2122.
Chakrabarti, D., Punera, K. (2011). Event Summarization Using Tweets. ICWSM, 11,
66-73.
Chen, L., Cong, G., Jensen, C. S., Wu, D. (2013). Spatial keyword query processing: An
experimental evaluation. In Proceedings of the VLDB Endowment (Vol. 6, No. 3, pp.
217-228). VLDB Endowment.
Chen, Rui, Raj Sharman, H. Raghav Rao, and Shambhu J. Upadhyaya (2008).
Coordination in emergency response management. Commun. ACM 51, 5 (May
2008), 66-73.
Chen, W. P., Hou, J. C., Sha, L. (2004). Dynamic clustering for acoustic target tracking
in wireless sensor networks. Mobile Computing, IEEE Transactions on, 3(3), 258-271.
Chen, Y. Y., Suel, T., Markowetz, A. (2006). Efficient query processing in geographic
web search engines. In Proceedings of the 2006 ACM SIGMOD international
conference on Management of data (pp. 277-288). ACM.
Chen, Z., Zhang, J., Li, Z., Wu, F., Ho, K. (2008). The technical concept within the
Integrative landslide early warning system. In Proc. of Int’l. Symposium on Landslides
and Engineered Slopes (pp. 1083-1088).
104
City.Risks
Deliverable D2.1
Choi, D. W., Chung, C. W., Tao, Y. (2014). Maximizing Range Sum in External
Memory. ACM Transactions on Database Systems (TODS), 39(3), 21.
Chow, C. Y., Mokbel, M. F. (2011). Trajectory privacy in location-based services and
data publication. ACM SIGKDD Explorations Newsletter, 13(1), 19-29.
Christoforaki, M., He, J., Dimopoulos, C., Markowetz, A., Suel, T. (2011). Text vs.
space: efficient geo-search query processing. In Proceedings of the 20th ACM
international conference on Information and knowledge management (pp. 423-432).
ACM.
Chun, Wendy H., Höllerer, Tobias (2013). Real-time hand interaction for augmented
reality on mobile phones. In: Proceedings of the 2013 international conference on
Intelligent user interfaces (IUI '13). ACM, New York, NY, USA, 307-314.
Cinque, M., Cotroneo, D., Pecchia, A. (2013). Event logs for the analysis of software
failures: A rule-based approach. Software Engineering, IEEE Transactions on, 39(6),
806-821.
Cisco
(2015),
Cisco
Smart+Connected
City
Operations
http://www.cisco.com/web/strategy/smart_connected_communities/cityoperations-center.html (Accessed: 20. July 2015).
Center
Cong, G., Jensen, C. S., Wu, D. (2009). Efficient retrieval of the top-k most relevant
spatial web objects. Proceedings of the VLDB Endowment, 2(1), 337-348.
Cong, G., Lu, H., Ooi, B. C., Zhang, D., Zhang, M. (2012). Efficient spatial keyword
search in trajectory databases. arXiv preprint arXiv:1205.2880.
Cross, N. (1992). Research in Design Thinking, in Research in Design Thinking, N.
Cross, K. Dorst and N. Roozenburg (eds.). Delft, Netherlands: Delft University Press.
CycleLeash
(2014).
Case
Study,
content/uploads/2014/12/CycleLeash-Case-Study.pdf
http://bluvision.com/wp-
Dave, K., Lawrence, S., Pennock, D. M. (2003). Mining the peanut gallery: Opinion
extraction and semantic classification of product reviews. In Proceedings of the 12th
international conference on World Wide Web (pp. 519-528). ACM.
De Almeida, V. T., Güting, R. H. (2005). Indexing the trajectories of moving objects in
networks. GeoInformatica, 9(1), 33-60.
Dean, J., , S. (2008). MapReduce: simplified data processing on large clusters.
Communications of the ACM, 51(1), 107-113.
Delling, D., Wagner, D. (2007). Landmark-based routing in dynamic graphs. In
Experimental algorithms (pp. 52-65). Springer Berlin Heidelberg.
Dementyev, A., Hodges, S., Taylor, S., Smith, J. (2013, April). Power consumption
analysis of Bluetooth Low Energy, ZigBee and ANT sensor nodes in a cyclic sleep
scenario. In Wireless Symposium (IWS), 2013 IEEE International (pp. 1-4). IEEE.
Dickson, E. (2012) Urban risk assessments: understanding disaster and climate risk in
cities. World Bank Publications.
105
City.Risks
Deliverable D2.1
Dong, Q., Yu, L., Lu, H., Hong, Z., Chen, Y. (2010). Design of building monitoring
systems based on wireless sensor networks. Wireless Sensor Network, 2(09), 703.
D’Angelo, G., D’Emidio, M., Frigioni, D., Vitale, C. (2012). Fully dynamic maintenance
of arc-flags in road networks. In Experimental Algorithms (pp. 135-147). Springer
Berlin Heidelberg.
Eck, J., Chainey, S., Cameron, J., Wilson, R. (2005) Mapping crime: Understanding
hotspots.
Ergen, S. ZigBee/IEEE (2004). 802.15.4 Summary, UC, Berkeley, Calif, USA.
Etzion, O., Niblett, P. (2010). Event processing in action. Manning Publications Co..
Farahani, R. Z., Asgari, N., Heidari, N., Hosseininia, M., Goh, M. (2012). Covering
problems in facility location: A review. Computers Industrial Engineering, 62(1), 368407.
Ferrari, L., Rosi, A., Mamei, M., Zambonelli, F. (2011). Extracting urban patterns from
location-based social networks. In Proceedings of the 3rd ACM SIGSPATIAL
International Workshop on Location-Based Social Networks (pp. 9-16). ACM.
Florczyk, A. J, F. J López-Pellicer, J. Valino, R. Béjar and P. R. Muro-Medrano. (200).
INSPIRE-able Services. In: 12th AGILE International Conference on Geographic Information Science. Hannover, Germany.
French Scrum User Group (2009). A National Survey on Agile Methods in France.
www.frenchsug.org
Frentzos, E. (2003). Indexing objects moving on fixed networks. In Advances in
spatial and temporal databases (pp. 289-305). Springer Berlin Heidelberg.
Fu, K., Lu, Y. C., Lu, C. T. (2014). TREADS: a safe route recommender using social
media mining and text summarization. In Proceedings of the 22nd ACM SIGSPATIAL
International Conference on Advances in Geographic Information Systems (pp. 557560). ACM.
Fuchs-Kittowski, F., Simroth, S.; Pfützner, B., Trosien, F. (2012). Mobile Augmented
Maps und Aug-mented Reality für den präventiven Hochwasserschutz. In: Knetsch,
G. (Hrsg.): Umweltinformati-onssysteme – Frühwarn- und Informationssysteme für
den Hochwasserschutz. UBA-Texte Nr. 41/2012, Dessau: Umweltbundesamt, 19-32.
Fuchs-Kittowski, Frank, Faust, Daniel (2014). Architecture of Mobile Crowdsourcing
Systems. In: Pro-ceedings of the 20th International Conference on Collaboration and
Technology (CRIWG 2014), Santiago, Chile, September 7th-10th 2014, Springer
International Publishing, Berlin, Heidelberg, New York, 129-144.
Furuhata, M., Dessouky, M., Ordóñez, F., Brunet, M. E., Wang, X., Koenig, S. (2013).
Ridesharing: The state-of-the-art and future directions. Transportation Research Part
B: Methodological, 57, 28-46.
Ganti, R. K., Ye, F., Lei, H. (2011). Mobile crowdsensing: current state and future
challenges. Communications Magazine, IEEE, 49(11), 32-39.
106
City.Risks
Deliverable D2.1
Garcia, C., Ibarguengoytia-Gonzalez, P., Garcia-Hernandez, J., and Perez-Diaz, J.
(2007). Wireless sensor networks and applications: a survey, International Journal of
Computer Science and Network Security, vol. 7, no. 3, pp. 264–273.
García-Hernández, C. F., Ibarguengoytia-Gonzalez, P. H., García-Hernández, J., PérezDíaz, J. A. (2007). Wireless sensor networks and applications: a survey. IJCSNS
International Journal of Computer Science and Network Security, 7(3), 264-273.
Geisberger, R., Sanders, P., Schultes, D., Vetter, C. (2012). Exact routing in large road
networks using contraction hierarchies. Transportation Science, 46(3), 388-404.
Gerber, M. S. (2014). Predicting crime using Twitter and kernel density estimation.
Decision Support Systems, 61, 115-125.
Goldberg, A. V., Harrelson, C. (2005). Computing the shortest path: A search meets
graph theory. In Proceedings of the sixteenth annual ACM-SIAM symposium on
Discrete algorithms (pp. 156-165). Society for Industrial and Applied Mathematics.
Goldberg, A. V., Werneck, R. F. F. (2005). Computing Point-to-Point Shortest Paths
from External Memory. In ALENEX/ANALCO (pp. 26-40).
Gruteser, M., Hoh, B. (2005). On the anonymity of periodic location samples. In
Security in Pervasive Computing (pp. 179-192). Springer Berlin Heidelberg.
Gröger, Gerhard, Kolbe, Thomas H., Nagel, Claus, Häfele, Karl-Heinz (2012). CityGML
2.0 - OGC City Geography Markup Language (CityGML). Open Geospatial Consortium
(OGC 12-019), http://www.opengis.net/spec/citygml/2.0.
Gurpreet Singh Matharu, Anju Mishra, “Empirical Study of Agile Software
Development Methodologies: A Comparative Analysis”, ACM SIGSOFT Software
Engineering Notes, January 2015 Volume 40 Number 1
Ha, M., Kim, D., Kim, S. H. and Hong, S. (2010). Inter-MARIO: a fast and seamless
mobility protocol to support inter-PAN handover in 6LoWPAN’, in Proceedings of the
53rd IEEE Global Communications Conference (GLOBECOM 10).
Hamacher, H. W., Tjandra, S. A. (2001). Mathematical modelling of evacuation
problems: A state of art. Fraunhofer-Institut für Techno-und Wirtschaftsmathematik,
Fraunhofer (ITWM).
Hariharan, R., Hore, B., Li, C., Mehrotra, S. (2007). Processing spatial-keyword (SK)
queries in geographic information retrieval (GIR) systems. In Scientific and Statistical
Database Management, 2007. SSBDM'07. 19th International Conference on (pp. 1616). IEEE.
Harries, K.A. (1999) Mapping crime: Principle and practice.
Havlik, D., G. Schimak, R. Denzer and B. Stevenot. (2006) Introduction to SANY
(sensors anywhere) integrated project. Shaker, Graz, Austria: 541–546.
He, T., Krishnamurthy, S., Stankovic, J. A., Abdelzaher, T., Luo, L., Stoleru, R., ...
Krogh, B. (2004, June). Energy-efficient surveillance system using wireless sensor
networks. In Proceedings of the 2nd international conference on Mobile systems,
applications, and services (pp. 270-283). ACM.
107
City.Risks
Deliverable D2.1
Hilger, M., Köhler, E., Möhring, R. H., Schilling, H. (2009). Fast point-to-point shortest
path computations with arc-flags. The Shortest Path Problem: Ninth DIMACS
Implementation Challenge, 74, 41-72.
Hoh, B., Gruteser, M., Xiong, H., Alrabady, A. (2010). Achieving guaranteed
anonymity in gps traces via uncertainty-aware path cloaking. Mobile Computing,
IEEE Transactions on, 9(8), 1089-1107.
Hollifield, B.R., Habibi, E. (2006) The Alarm Management Handbook: A
Comprehensive Guide: Practical and Proven Methods to Optimize the Performance
of Any Alarm Management System. Pas.
Holma, H. and Toskala, A. (2006). HSDPA/HSUPA for UMTS: High Speed Radio Access
for Mobile Communications, John Wiley Sons, Chichester, England.
Hong, S., Kim, D., Ha, M. et al (2010). SNAIL: an IP-based wireless sensor network
approach to the Internet of things’, IEEE Wireless Communications, vol. 17, no. 6, pp.
34–42.
Huang, Zhanpeng, Li, Weikai, Hui, Pan, Peylo, Christoph (2014). CloudRidAR - a cloudbased architecture for mobile augmented reality. In: Proceedings of the 2014
workshop on Mobile augmented reality and robotic technology-based systems
(MARS '14). ACM, New York, NY, USA, 29-34.
IBM (2012) Intelligent Operations Center for Smarter Cities Administration Guide.
IBM
Corporation,
International
Technical
Support
Organization
http://www.redbooks.ibm.com/redbooks/pdfs/sg248061.pdf (Accessed: 20. July
2015).
ITU Internet Report (2005). The internet of things’.
Itria, M. L., Daidone, A., Ceccarelli, A. (2014). A Complex Event Processing Approach
for Crisis-Management Systems. arXiv preprint arXiv:1404.7551.
Jagtman, H. M. (2010). Cell broadcast trials in the Netherlands: Using mobile phone
technology for citizens alarming, Reliability Engineering and System Safety, vol. 95
no. 1, pp. 18-28, 2010.
Jain, S., McLean, C. R. (2004). An architecture for integrated modeling and
simulation for emergency response. In Proceedings of the institute of industrial
engineers annual conference (pp. 1503-1508).
Jensen, C. S., Lu, H., Yang, B. (2009). Indexing the trajectories of moving objects in
symbolic indoor space. In Advances in Spatial and Temporal Databases (pp. 208-227).
Springer Berlin Heidelberg.
Kallus, N. (2014). Predicting crowd behavior with big public data. In Proceedings of
the companion publication of the 23rd international conference on World wide web
companion (pp. 625-630). International World Wide Web Conferences Steering
Committee.
Kapucu, Naim (2005). Interorganizational coordination in dynamic context: Networks
in emergency response management. Connections 26.2 (2005): 33-48.
108
City.Risks
Deliverable D2.1
Kapucu, Naim. (2008). Collaborative emergency management: better community
organising, better public preparedness and response. Disasters, 32: 239–262
Kido, H., Yanagisawa, Y., Satoh, T. (2005). An anonymous communication technique
using dummies for location-based services. In Pervasive Services, 2005. ICPS'05.
Proceedings. International Conference on (pp. 88-97). IEEE.
Kim, M.-K. and Jee, K.-Y. (2006). Characteristics of individuals influencing adoption
intentions for portable internet service’, ETRI Journal, vol. 28, no. 1, pp. 67–76.
Kim, S., George, B., Shekhar, S. (2007). Evacuation route planning: scalable
heuristics. In Proceedings of the 15th annual ACM international symposium on
Advances in geographic information systems (p. 20). ACM.
Kim, Y. J., Hong, S., Lee, J. U., Song, S., Kim, D. (2013). An integration avenue of
ground monitoring based on wireless sensor networks. International Journal of
Distributed Sensor Networks, 2013.
- Actuatorbased Early Warning System for Extreme Weather Conditions. In: Pebesma, E.; Bishr,
M.; Bartoschek, T. (eds.): Proceedings of the International 6th Geographic
Information Da
-330, 2008.
Klafft, Michael (2013). Diffusion of emergency warnings via multi-channel
communication systems an empirical analysis, Autonomous Decentralized Systems
(ISADS), 2013 IEEE Eleventh International Symposium on , vol., no., pp.1,5, 6-8 March
2013
Kling, F., Pozdnoukhov, A. (2012). When a city tells a story: urban topic analysis. In
Proceedings of the 20th International Conference on Advances in Geographic
Information Systems (pp. 482-485). ACM.
Kriegel, H. P., Renz, M., Schubert, M. (2010). Route skyline queries: A multipreference path planning approach. In Data Engineering (ICDE), 2010 IEEE 26th
International Conference on (pp. 261-272). IEEE.
Kuorilehto, M., Hännikäinen, M., Hämäläinen, T. D. (2005). A survey of application
distribution in wireless sensor networks. EURASIP Journal on Wireless
Communications and Networking, 2005(5), 774-788.
Kwak, H., Lee, C., Park, H., Moon, S. (2010). What is Twitter, a social network or a
news media?. In Proceedings of the 19th international conference on World wide
web (pp. 591-600). ACM.
Lai, T. K., Wang, A., Chang, C. M., Tseng, H. M., Huang, K., Li, J. P., ... Chou, P. H.
(2014, April). Demonstration abstract: an 8x8 mm 2 bluetooth low energy wireless
motion-sensing platform. In Proceedings of the 13th international symposium on
Information processing in sensor networks (pp. 341-342). IEEE Press.
Lai, T., Lin, C., Su, Y., Chu, H. (2011). BikeTrack: Tracking stolen bikes through
everyday mobile phones and participatory sensing. ACM PhoneSense.
Lamprianidis, G., Skoutas, D., Papatheodorou, G., Pfoser, D. (2014). Extraction,
integration and analysis of crowdsourced points of interest from multiple web
109
City.Risks
Deliverable D2.1
sources. In Proceedings of the 3rd ACM SIGSPATIAL International Workshop on
Crowdsourced and Volunteered Geographic Information (pp. 16-23). ACM.
Lansdall-Welfare, T., Lampos, V., Cristianini, N. (2012). Effects of the Recession on
Public Mood in the UK. In Proceedings of the 21st international conference
companion on World Wide Web (pp. 1221-1226). ACM.
Laptev, D., Tikhonov, A., Serdyukov, P., Gusev, G. (2014). Parameter-free discovery
and recommendation of areas-of-interest. In Proceedings of the 22nd ACM
SIGSPATIAL International Conference on Advances in Geographic Information
Systems (pp. 113-122). ACM.
Le, T. T. T., Nickerson, B. G. (2008). Efficient search of moving objects on a planar
graph. In Proceedings of the 16th ACM SIGSPATIAL international conference on
Advances in geographic information systems (p. 41). ACM.
LeanKit (2014). Maximize Your Time, Improve Efficiency with the Kanban System.
http://leankit.com/kanban/kanban-system/
Lechner, Martin (2013). OGC Augmented Reality Markup Language 2.0 (ARML 2.0).
OGC Draft Implementation Specification (OGC 12-132r2), Open Geospatial
Consortium (OGC), http://www.opengis.net/doc/arml2x0/2.0.
Li, D., Wong, K. D., Hu, Y. H., Sayeed, A. M. (2002). Detection, classification, and
tracking of targets. Signal Processing Magazine, IEEE, 19(2), 17-29.
Li, L., Halpern, J. Y., Bahl, P., Wang, Y.-M. and Wattenhofer, R. (2001). Analysis of a
cone-based distributed topology control algorithm for wireless multi-hop networks’,
In Proceedings of the ACM Symposium on Principles of Distributed Computing
(PODC), pp. 264–273.
Li, N. and Hou, J. C. (2004). FLSS: A fault-tolerant topology control algorithm for
wireless networks’, In Proceedings of the ACM International Conference on Mobile
Computing and Networking (MobiCom), Philadelphia, Pennsylvania.
Li, N., Hou, J. C. and Sha, L. (2003). Design and analysis of an MST-based topology
control algorithm’, In Proce of IEEE INFOCOM.
Li, R., Lei, K. H., Khadiwala, R., Chang, K. C. C. (2012). Tedas: A twitter-based event
detection and analysis system. In Data engineering (icde), 2012 ieee 28th
international conference on (pp. 1273-1276). IEEE.
Li, X., Stojmenovic, I. and Wang, Y. (2004). Partial Delaunay triangulation and degree
limited localized Bluetooth scatternet formation’, IEEE Transactions on Parallel and
Distributed Systems, vol. 15, no. 4, pp. 350–361.
Li, X., Zhao, Z., Zhu, X., Wyatt, T. (2011). Covering models and optimization
techniques for emergency response facility location and planning: a review.
Mathematical Methods of Operations Research, 74(3), 281-310.
Lieberman, M. D., Samet, H. (2012). Adaptive context features for toponym
resolution in streaming news. In Proceedings of the 35th international ACM SIGIR
conference on Research and development in information retrieval (pp. 731-740).
ACM.
110
City.Risks
Deliverable D2.1
Lim, G. J., Zangeneh, S., Baharnemati, M. R., Assavapokee, T. (2012). A capacitated
network flow optimization approach for short notice evacuation planning. European
Journal of Operational Research, 223(1), 234-245.
Lindh, Joakim (2015) SWRA475, TI-Application report, Bluetooth Low Energy beacons
Liu, H., Jia, X., Wan, P. J., Yi, C. W., Makki, S. K., Pissinou, N. (2007). Maximizing
lifetime of sensor surveillance systems. Networking, IEEE/ACM Transactions on,
15(2), 334-345.
Liu, J., Yu, G., Sun, H. (2011). Subject-oriented top-k hot region queries in spatial
dataset. In Proceedings of the 20th ACM international conference on Information
and knowledge management (pp. 2409-2412). ACM.
Luo, J., Joshi, D., Yu, J., Gallagher, A. (2011). Geotagging in multimedia and computer
vision—a survey. Multimedia Tools and Applications, 51(1), 187-211.
Machanavajjhala, A., Kifer, D., Gehrke, J., Venkitasubramaniam, M. (2007). ldiversity: Privacy beyond k-anonymity. ACM Transactions on Knowledge Discovery
from Data (TKDD), 1(1), 3.
Maher, M., Poon, J., and Boulanger, S. (1995). Formalizing Design Exploration as CoEvolution: A Combined Gene Approach, in: Preprints of the Second IFIP WG5.2
Workshop on Advances in Formal Design Methods for CAD, J.S.G.F. Sudweeks (ed.).
Key Centre of Design Computing: pp. 1-28.
Manweiler, Justin (2014). Cloud-based mobile augmented reality, pitfalls and
strategies in a realtime deployment. In: Proceedings of the 2014 workshop on
Mobile augmented reality and robotic technology-based systems (MARS '14). ACM,
New York, NY, USA, 21-21.
Mendoza, M., Poblete, B., Castillo, C. (2010). Twitter Under Crisis: Can we trust what
we RT?. In Proceedings of the first workshop on social media analytics (pp. 71-79).
ACM.
Metaio (2014). AREL - Augmented
http://dev.metaio.com/arel/overview/.
Reality
Experience
Language.
Mohler, G. O., Short, M. B., Brantingham, P. J., Schoenberg, F. P., Tita, G. E. (2012).
Self-exciting point process modeling of crime. Journal of the American Statistical
Association.
Mokbel, M. F., Chow, C. Y., Aref, W. G. (2006). The new Casper: query processing for
location services without compromising privacy. In Proceedings of the 32nd
international conference on Very large data bases (pp. 763-774). VLDB Endowment.
Mokbel, M. F., Ghanem, T. M., Aref, W. G. (2003). Spatio-temporal access methods.
IEEE Data Eng. Bull., 26(2), 40-49.
Murphy, B., et al.. (2013). Have Agile Techniques been the Silver Bullet for Software
Development at Microsoft?, Empirical Software Engineering and Measurement, ACM
/ IEEE International Symposium,(2013), 75-84. DOI=10.1109/ESEM.2013.21.
111
City.Risks
Deliverable D2.1
Murray-Tuite, P., Wolshon, B. (2013). Evacuation transportation modeling: An
overview of research, development, and practice. Transportation Research Part C:
Emerging Technologies, 27, 25-45.
Naphade, M., Banavar, G., Harrison, C., Paraszczak, J., Morris, R., (2011) Smarter
Cities and Their Innovation Challenges. Computer 44, 32–39.
Levine, Ned (2015). CrimeStat: A Spatial Statistics Program for the Analysis of Crime
Incident Locations (v 4.02). Ned Levine Associates, Houston, Texas, and the National
Institute of Justice, Washington, D.C. August.
Nepomnyachiy, S., Gelley, B., Jiang, W., Minkus, T. (2014). What, where, and when:
keyword search with spatio-temporal ranges. In Proceedings of the 8th Workshop on
Geographic Information Retrieval (p. 2). ACM.
Nergiz, M. E., Atzori, M., Saygin, Y. Güç, B. 2009. Towards trajectory anonymization:
A generalization-based approach. Trans. Data Privacy, 2(1), 47-75.
Next Generation Wireless Security Application Note (2015). March 2015, Siklu
Nguyen-Dinh, L. V., Aref, W. G.,
methods: Part 2 (2003-2010).
Mokbel, M. (2010). Spatio-temporal access
Ni, J., Ravishankar, C. V. (2005). PA-tree: A parametric indexing scheme for spatiotemporal trajectories. In Advances in Spatial and Temporal Databases (pp. 254-272).
Nilsson, Rolf (2015) Shaping the Wireless Future with Low Energy Applications and
Systems, ConnectBlue
Onodera, K. and Miyazaki, T. (2008). An Autonomous Algorithm for Construction of
Energy-conscious Communication Tree in Wireless Sensor Networks’, in Advanced
Information Networking and Applications - Workshops.
Perey, Christine (2014). AR Standards - AR Community for Open and Interoperable
Augmented Reality Experiences. http://www.perey.com/ARStandards/.
Perry, R. W. (2004). Disaster Exercise Outcomes for Professional Emergency
Personnel and Citizen Volunteers. Journal of Contingencies and Crisis Management,
12: 64–75.
Pillac, V., Gendreau, M., Guéret, C., Medaglia, A. L. (2013). A review of dynamic
vehicle routing problems. European Journal of Operational Research, 225(1), 1-11.
Ralph, P. (2015). The Sensemaking-Coevolution-Implementation Theory of Software
Design. Science of Computer Programming, 101, 21-41.
Ratcliffe, J. (2010). Crime mapping: spatial and temporal challenges. In Handbook of
quantitative criminology (pp. 5-24). Springer New York.
Rice, M. N., Tsotras, V. J. (2012). Exact graph search algorithms for generalized
traveling salesman path problems. In Experimental Algorithms (pp. 344-355).
Rice, M. N., Tsotras, V. J. (2013). Parameterized algorithms for generalized traveling
salesman problems in road networks. In Proceedings of the 21st ACM SIGSPATIAL
112
City.Risks
Deliverable D2.1
International Conference on Advances in Geographic Information Systems (pp. 114123). ACM.
Richter, K. F., Shi, M., Gan, H. S., Winter, S. (2013). Decentralized evacuation
management. Transportation Research Part C: Emerging Technologies, 31, 1-17.
Rocha-Junior, J. B., Nørvåg, K. (2012). Top-k spatial keyword queries on road
networks. In Proceedings of the 15th international conference on extending
database technology (pp. 168-179). ACM.
Royce, W.W. (1970). Managing the Development of Large Software Systems:
Concepts and Techniques, Proceedings of Wescon.
Rumiński, Dariusz, Walczak, Krzysztof (2013). Creation of Interactive AR Content on
Mobile Devices , In: Wil van der Aalst, John Mylopoulos, Michael Rosemann, M. J.
Shaw, and C. Szyperski (eds.), International Conference on Business Information
Systems, Poznań, Poland, 19 - 20 June, 2013 , Lecture Notes in Business Information
Processing, vol. 160, ed., Springer, Berlin-Heidelberg, 258-269.
Rumiński, Dariusz, Walczak, Krzysztof (2014). CARL - A Language for Modelling
Contextual Augmented Reality Environments. In: Camarinha-Matos, L. M., N. S.
Barrento, and R. Mendonça (eds.): Technological Innovation for Collective
Awareness Systems, Springer, Berlin-Heidelberg, 183-190.
SRC WG (2013). X3D AR Proposal (Public Review). Augmented Reality Continuum
Working
Group
(ARC
WG),
Web3D
Consortium,
http://www.web3d.org/wiki/index.php?title=AR_Proposal_Public_Review.
Sakaki, T., Okazaki, M., Matsuo, Y. (2010). Earthquake shakes Twitter users: realtime event detection by social sensors. In Proceedings of the 19th international
conference on World wide web (pp. 851-860). ACM.
Salo, O. and Abrahamsson, P. (2008). Agile Methods in European Embedded
Development Organizations: a survey study of Extreme Programming and Scrum, IET
Software, vol. 2 (February, 2008), 58-64. DOI = 10.1049/iet-sen:20070038
Samarati, P. (2001). Protecting respondents identities in microdata release.
Knowledge and Data Engineering, IEEE Transactions on, 13(6), 1010-1027.
Sanders, P., Schultes, D., Vetter, C. (2008). Mobile route planning. In Algorithms-ESA
2008 (pp. 732-743). Springer Berlin Heidelberg.
Sankaranarayanan, J., Samet, H., Teitler, B. E., Lieberman, M. D., Sperling, J. (2009).
Twitterstand: news in tweets. In Proceedings of the 17th acm sigspatial international
conference on advances in geographic information systems (pp. 42-51). ACM.
Sassen, A.-M., Annoni, A., Millot, M., Denzer, R., Hecht, L., Pichler, G., Couturier, M.,
Alegre, C., (2005) Towards an Open Risk Management Service Architecture, in:
Proceedings of the Second International ISCRAM Conference.
Schach, S. (2007). Software Engineering, Tata McGraw Hill, Ed. 7, 4-6.
Semiocast, (2012). Twitter reaches half a billion accounts. More than 140 millions in
the U.S., http://semiocast.com/publications/2012_07_30_Twitter_reaches_half_a_
billion_accounts_140m_in_the_US, October 21, 2013.
113
City.Risks
Deliverable D2.1
Serdyukov, P., Murdock, V., Van Zwol, R. (2009). Placing flickr photos on a map. In
Proceedings of the 32nd international ACM SIGIR conference on Research and
development in information retrieval (pp. 484-491). ACM.
Sharifi, M., Okhovvat, M. (2012). Scate: A scalable time and energy aware actor task
allocation algorithm in wireless sensor and actor networks. ETRI Journal, 34(3), 330340.
Shi, P.J., Du, J., Ji, M., Liu, J., Wang, J. (2006) Urban risk assessment research of major
natural disasters in China. Advances in Earth Science 21, 170–177.
Skovsgaard, A., Sidlauskas, D., Jensen, C. S. (2014). Scalable top-k spatio-temporal
term querying. In Data Engineering (ICDE), 2014 IEEE 30th International Conference
on (pp. 148-159). IEEE.
Starbird, Kate and Leysia Palen (2011). Voluntweeters: self-organizing by digital
volunteers in times of crisis. In Proceedings of the SIGCHI Conference on Human
Factors in Computing Systems (CHI '11). ACM, New York, NY, USA, 1071-1080.
Stojmenovic, I. (2005). Handbook of Sensor Networks: Algorithms and Architectures,
John Wiley Sons, Chichester, England.
Strapparava, C., Valitutti, A. (2004). WordNet Affect: an Affective Extension of
WordNet. In LREC (Vol. 4, pp. 1083-1086).
Sweeney, L. (2002). k-anonymity: A model for protecting privacy. International
Journal of Uncertainty, Fuzziness and Knowledge-Based Systems, 10(05), 557-570.
Tan, R., Xing, G., Wang, J., So, H. C. (2008, June). Collaborative target detection in
wireless sensor networks with reactive mobility. In Quality of Service, 2008. IWQoS
2008. 16th International Workshop on (pp. 150-159). IEEE.
Teitler, B. E., Lieberman, M. D., Panozzo, D., Sankaranarayanan, J., Samet, H.,
Sperling, J. (2008). NewsStand: A new view on news. In Proceedings of the 16th ACM
SIGSPATIAL international conference on Advances in geographic information systems
(p. 18). ACM.
Texas Instruments (2007). 2.4 GHz IEEE 802.15.4 / ZigBee-Ready RF Transceiver,
http://focus.ti.com/lit/ds/symlink/cc2420.pdf.
Tozlu, S., Senel, M., Mao, W., Keshavarzian, A. (2012). Wi-Fi enabled sensors for
internet of things: A practical approach. Communications Magazine, IEEE, 50(6), 134143.
Tyshchuk, Y., Hui, C., Grabowski, M., Wallace, W.A. (2012). Social Media and Warning
Response Impacts in Extreme Events: Results from a Naturally Occurring Experiment,
System Science (HICSS), 2012 45th Hawaii International Conference on, vol., no.,
pp.818,827, 4-7 Jan. 2012
United Nations (2009). UN-ISDR Terminology on Disaster Risk Reduction,
International Strategy for Disaster Reduction (ISDR), UN-ISDR Secretariat, Geneva.
http://www.unisdr.org/files/7817_UNISDRTerminologyEnglish.pdf (Accessed: 20.
July 2015).
114
City.Risks
Deliverable D2.1
Vaid, S., Jones, C. B., Joho, H., Sanderson, M. (2005). Spatio-textual indexing for
geographical search on the web. In Advances in Spatial and Temporal Databases (pp.
218-235). Springer Berlin Heidelberg.
Van de Ven, A.H., and Poole, M.S. (1995). Explaining Development and Change in
Organizations, The Academy of Management Review (20:3), July, pp. 510-540.
Vansteenwegen, P., Souffriau, W., Van Oudheusden, D. (2011). The orienteering
problem: A survey. European Journal of Operational Research, 209(1), 1-10.
Versionone
(2013).
8th
Annual
State
of
http://www.versionone.com/pdf/2013-state-of-agile-survey.pdf
Agile
Survey,
Vieweg, Sarah, Amanda L. Hughes, Kate Starbird, and Leysia Palen (2010).
Microblogging during two natural hazards events: what twitter may contribute to
situational awareness. In Proceedings of the SIGCHI Conference on Human Factors in
Computing Systems (CHI '10). ACM, New York, NY, USA, 1079-1088.
Vlugt, Erik (2013) Bluetooth Low Energy, Beacons and Retail, Verifone 2013
Wang, L., Zheng, Y., Xie, X., Ma, W. Y. (2008). A flexible spatio-temporal indexing
scheme for large-scale GPS track retrieval. In Mobile Data Management, 2008.
MDM'08. 9th International Conference on (pp. 1-8). IEEE.
Wang, X., Gerber, M. S., Brown, D. E. (2012). Automatic crime prediction using
events extracted from twitter posts. In Social Computing, Behavioral-Cultural
Modeling and Prediction (pp. 231-238). Springer Berlin Heidelberg.
Wang,X. S., Zhan, Y. Z. and Wang, L. (2008). STCP: Secure Topology Control Protocol
for WSNs Based on Hexagonal Mesh’, Wireless Comm, Networking and Mobile
Computing.
Washburn, D., Sindhu, U., Balaouras, S., Dines, R. A., Hayes, N., Nelson, L. E. (2009).
Helping CIOs understand “smart city” initiatives. Growth, 17.
Weng, J., Lee, B. S. (2011). Event Detection in Twitter. ICWSM, 11, 401-408.
Wheeler, A. (2007). Commercial applications of wireless sensor networks using
ZigBee’, IEEE Communications Magazine, vol. 45, no. 4, pp. 70–77.
Wing, B. P., Baldridge, J. (2011). Simple supervised document geolocation with
geodesic grids. In Proceedings of the 49th Annual Meeting of the Association for
Computational Linguistics: Human Language Technologies-Volume 1 (pp. 955-964).
Association for Computational Linguistics.
Winter, S., Iglesias, G., Zlatanova, S., Kuhn, W. (2005). GI for the public: the terrorist
attack in London. Geoinformatics Magazine, (7), 18-21.
Wu, D., Cong, G., Jensen, C. S. (2012). A framework for efficient spatial web object
retrieval. The VLDB Journal—The International Journal on Very Large Data Bases,
21(6), 797-822.
Wu, D., Yiu, M. L., Jensen, C. S. (2013). Moving spatial keyword queries:
Formulation, methods, and analysis. ACM Transactions on Database Systems (TODS),
38(1), 7.
115
City.Risks
Deliverable D2.1
Xiao, X., Yi, K., Tao, Y. (2010). The hardness and approximation algorithms for ldiversity. In Proceedings of the 13th International Conference on Extending
Database Technology (pp. 135-146). ACM.
Xue, Y., Brown, D. E. (2006). Spatial analysis with preference specification of latent
decision makers for criminal event prediction. Decision support systems, 41(3), 560573.
Yao, B., Tang, M., Li, F. (2011). Multi-approximate-keyword routing in gis data. In
Proceedings of the 19th ACM SIGSPATIAL International Conference on Advances in
Geographic Information Systems (pp. 201-210). ACM.
Yeon, S. J. (2007). A strategic grid for implementing ubiquitous computing. ETRI
journal, 29(2), 252-254.
Yick, J., Mukherjee, B., Ghosal, D. (2008). Wireless sensor network survey. Computer
networks, 52(12), 2292-2330.
Yuan, Yufei and Brian Detlor (2005). Intelligent mobile crisis response systems.
Commun. ACM 48, 2 (February 2005), 95-98.
Zaharia, M., Chowdhury, M., Franklin, M. J., Shenker, S., Stoica, I. (2010, June).
Spark: cluster computing with working sets. In Proceedings of the 2nd USENIX
conference on Hot topics in cloud computing (Vol. 10, p. 10).
Zhang, D., Chee, Y. M., Mondal, A., Tung, A. K., Kitsuregawa, M. (2009). Keyword
search in spatial databases: Towards searching by document. In Data Engineering,
2009. ICDE'09. IEEE 25th International Conference on (pp. 688-699). IEEE.
Zhao, C., Zhou, Z. and Kwak, K. (2010). Novel UWB transceiver for WBAN networks: a
study on AWGN channels’, ETRI Journal, vol. 32, no. 1, pp. 11–21.
Zhao, F., Shin, J., Reich, J. (2002). Information-driven dynamic sensor collaboration
for tracking applications,[IEEE Signal Process.
Zhou, P., Zhang, D., Salzberg, B., Cooperman, G., Kollios, G. (2005). Close pair
queries in moving object databases. In Proceedings of the 13th annual ACM
international workshop on Geographic information systems (pp. 2-11). ACM.
Zhou, Y., Xie, X., Wang, C., Gong, Y., Ma, W. Y. (2005). Hybrid index structures for
location-based web search. In Proceedings of the 14th ACM international conference
on Information and knowledge management (pp. 155-162). ACM.
ZigBee Alliance, http://www.zigbee.org/.
Zlatanova, S. (2005). Crisis designs. Geospatial Today, 4(1), 30-36.
Zlatanova, S., Holweg, D. (2004). 3D Geo-information in emergency response: a
framework. In Proceedings of the Fourth International Symposium on Mobile
Mapping Technology, Kunming, China (pp. 29-31).
116

Gaps and challenges for addressing security threats in urban

Transcription

Similar documents

Step 1. How to generate authorization code for your Supra EKEY

Alphonse Bertillion - Hicksville Public Schools / Homepage

iSCOUT PRS80

your seventh sense - L

ASSURANCE - Sensaphone

Answers the call, answers to you.

centrespread centrespread

Wireless Sensor Troubleshooting Guide

Skyline CPV36 Japanese Translation

a PDF copy here