MoteFinder: A Deployment Tool for Sensor Networks
Transcription
MoteFinder: A Deployment Tool for Sensor Networks
Rheinische Friedrich-Wilhelms-Universität Bonn Institute for Computer Science IV Sensor Networks and Pervasive Computing Group Römerstraße 164 D-53117 Bonn MoteFinder: A Deployment Tool for Sensor Networks Olga Saukh1, Robert Sauter1, Jonas Meyer2 and Pedro José Marrón1 1Universität Bonn, Bonn, Germany and Fraunhofer IAIS, St. Augustin, Germany 2Structural Engineering Research Laboratory, EMPA, Switzerland {saukh, sauter, pjmarron}@cs.uni-bonn.de, [email protected] Overview Motivation MoteFinder ◦ Hardware ◦ Software Performance Measurements ◦ Indoor Experiments ◦ Outdoor Experiments ◦ Advantages and Limitations Related Work Conclusions Sensor Networks and Pervasive Computing Group 2 Motivation Many applications of wireless sensor networks ◦ Little deployment experience ◦ Many unexpected problems in the deployment phase “Smart-dust” as a nice vision … ◦ but we are still far away from the use of “throw-away” sensor nodes ▪ Sensor nodes are (still) rather expensive ▪ But already hard to find! ▪ Throwing away hardware and batteries is ecologically not acceptable ▪ Reuse of sensor nodes is possible Tools for the deployment phase are required ◦ MoteFinder: A Deployment Tool for Sensor Networks Sensor Networks and Pervasive Computing Group 3 MoteFinder Is based on changing the omni-directional radiation/reception pattern of a standard sensor node antenna to a narrow beam ◦ A Directed antenna that is cheap and easy to build! Sensor Networks and Pervasive Computing Group 4 MoteFinder – Applications Selective interaction with one node or a group in the deployment phase ◦ Assignment of groups ◦ Node/Group configuration (position, …) ◦ Functionality check ◦ Measurement of the characteristics (e.g. TX-power) of individual sensor nodes and performance assessment Automatic or semi-automatic localization and collecting sensor nodes after an experiment ◦ ◦ ◦ ◦ Can be combined with robotics Environmental (and economic) considerations Finding malfunctioning nodes Tracking down of malicious sensor nodes that disturb the operation of the network Sensor Networks and Pervasive Computing Group 5 MoteFinder – Hardware Cantenna Tinfoil Cylinder + Sensor Networks and Pervasive Computing Group + 6 MoteFinder – Hardware Properties Cost: a few Euros Cost: a few Cents Time to build: < half a day Time to build: < half an hour ◦ Measurements, drilling Reduces energy efficiency Mote modification necessary Sensor Networks and Pervasive Computing Group 7 MoteFinder – Software LED blink frequency Sensor nodes with a LCD display TinyOS Oscilloscope Sensor Networks and Pervasive Computing Group 8 PDA display Evaluation Criteria Proof of Concept ◦ Experiments with both Cantenna and Tinfoil Cylinder ◦ Various settings for the Transmission Power Level (TPL) ◦ Various distances Indoor and outdoor scenarios 2D and 3D evaluated ◦ Different orientations of sensor nodes ◦ Different 3D placement of sensor nodes ◦ Different 3D orientation of the MoteFinder Sensor Networks and Pervasive Computing Group 9 Performance Measurements – Settings 32 markers of direction used to measure angles Angle between two neighboring markers is 11.2o 5s time difference between angle change 160s one full circle All coordinates in cm Z XY-Rotation Receiver Beacons MoteFinder YZ-Rotation (0,0,0) Ground ZX-Rotation Y Sensor Networks and Pervasive Computing Group 10 X Close-by measurement Polar coordinates Recorded RSSI values Actual direction of beacons (“mote”) shown Cantenna Distance: 40cm TPL: 2 Two measurements: ◦ same node & same settings Sensor Networks and Pervasive Computing Group 11 Indoor Setup 4 nodes located in different rooms of an apartment Walls: made of brick, 42cm thick (old building) mote 23 (350,520,110) mote 26 (-110,335,140) Beacons: TPL=5 (out of 31 for a Tmote Sky) MoteFinder (0,0,50) Different orientation of sensor nodes Different 3D locations of nodes Sensor Networks and Pervasive Computing Group mote 25 (510,-310,70) 12 mote 24 (108,-415,122) 0 degrees Indoor Results Cantenna Sensor Networks and Pervasive Computing Group Tinfoil Cylinder 13 Outdoor Setup mote 2 (0,300,0) mote 3 (310,20,0) MoteFinder (0,0,31) mote 1 (-150,-70,0) 3 beacon nodes located outdoors Beacons: TPL=31 Sensor Networks and Pervasive Computing Group 14 Outdoor Results Cantenna Sensor Networks and Pervasive Computing Group Tinfoil Cylinder 15 Evaluation Results (1) Proof of Concept Both Cantenna and Tinfoil Cylinder can be used ◦ Cantenna performs smoother with changing angle ◦ Cantenna outperforms Tinfoil Cylinder with reasonable TPL settings both indoors and outdoors ◦ Very rough error estimate: Cantenna ±30o, Tinfoil Cylinder ±45-60o Different values of Transmission Power Level ◦ Results show sensitivity to the TPL used by the beacon nodes ◦ Tinfoil Cylinder gives better results than Cantenna when used with maximum TPL of beacons at short distances ▪ Due to higher gain of Cantenna, which could result in the saturation of the input amplifier of the radio chip Different distances ◦ Due to higher gain of Cantenna, it can be successfully used at considerably larger distances than Tinfoil Cylinder Sensor Networks and Pervasive Computing Group 16 Evaluation Results (2) Indoor and outdoor scenarios ◦ Absence of obstacles reduces the error of direction estimates ◦ Tinfoil Cylinder performs smoother outdoors than indoors 2D and 3D support ◦ Different orientation of sensor nodes ▪ Slight difference in RSSI depending on the orientation of sensor nodes ▪ Negligible compared to changes in RSSI when rotating MoteFinder ◦ Different 3D placement of sensor nodes and 3D orientation of MoteFinder ▪ Difference in 3D location of beacons and MoteFinder can slightly increase the error of direction estimates Sensor Networks and Pervasive Computing Group 17 Advantages and Limitations Advantages: ◦ Wide applicability: ▪ MoteFinder can be successfully used for indoor and outdoor scenarios for both 2D and 3D settings ◦ The exactness of the results depends on ▪ Distance, presence of obstacles, TPL, MoteFinder itself Limitations: ◦ Only correctly formed messages are evaluated ◦ All beacons must include their IDs in the beacon messages to distinguish nodes Sensor Networks and Pervasive Computing Group 18 Related Work Cantennas ◦ Several works consider Cantennas for Wi-Fi networks [Chebrolu et. al, 2006] ◦ Used for extending range, not for localization [Johnson, 2007] [Beyers, 2002] ◦ Rural areas and underdeveloped countries Directional Antennas in Sensor Networks [Cho et. al, 2006] ◦ Longer lifetime by extending range of sink [Nasipuri et. al, 2002] [Rong et. al, 2006] ◦ Angle-of-arrival localization “Embrace and extend” ◦ Motes with LCD Displays, e.g., for deployment validation [Liu et. al, 2007] [Kotay et. al, 2005] ◦ Equip robots with directional antennas for guidance Sensor Networks and Pervasive Computing Group 19 Conclusions Cantenna and the Tinfoil cylinder work! ◦ Indoors and outdoors TPL/Antenna sensibility should be adjusted to distance for optimum performance Simple solution with lots of applications ◦ Deployment still difficult Helpful tools necessary ◦ Finding motes ◦ Selective interaction with groups/single nodes Future work: Combine with sensors Sensor Networks and Pervasive Computing Group 20 Thank you for Your Attention ? Are there any questions? + Robert Sauter [email protected] Universität Bonn Institut für Informatik IV Sensor Networks and Pervasive Computing Group Römerstr. 164 D-53117 Bonn Germany Sensor Networks and Pervasive Computing Group 21 +