AXE 10 - System Description
Transcription
AXE 10 - System Description
ERICSSON REVIEW PRESENTATION OF AXE 10 SWITCHING SYSTEM THE AXE 10 SYSTEM IN THE TELEPHONE NETWORK AXE 10-SYSTEM DESCRIPTION AXE 10-SOFTWARE STRUCTURE AND FEATURES 2 1976 NEW PACKAGING STRUCTURE FOR ELECTRONIC SWITCHING EQUIPMENT ERICSSON REVIEW NUMBER 2 • 1976 • VOLUME 53 Copyright Telefonaktiebolaget LM Ericsson Printed in Sweden, Stockholm 1976 RESPONSIBLE PUBLISHER EDITOR GUSTAF 0. EDITORIAL DR. T E C H N . C H R I S T I A N DOUGLAS STAFF FOLKE BERG EDITOR'S OFFICE S - 1 2 6 2 5 SUBSCRIPTION JACOB^EUS ONE YEAR STOCKHOLM $6.00, ONE COPY $1.70 Contents 54 • Presentation of AXE 10 Switching System 60 • Introduction of the AXE 10 Switching System in the Telephone Network 70 • AXE 10 — System Description 90 • AXE 10 — Software Structure and Features 100 • New Packaging Structure for Electronic Switching Equipment COVER AXE reed switch. Part of a printed-circuil-mounled switch board photographed on a mirror. The integrated circuits shown along the edge of the switch board contain control logic, thyristors and transistors for operating and holding of the reed crosspoints. Presentation of AXE 10 Switching System John Meurling As an introduction to the subsequent articles on AXE this paper provides a survey in the form of an outline discussion of the main system philosophy, the system structure and a presentation of the most important characteristics. UDC 621.395 722. 681.3.065 LME 83022 General structure AXE 10 is a stored program control telephone switching system designed for a full range of applications for local exchanges, tandem exchanges and small- to medium-size transit exchanges. The switching network is composed of reed relay switch-blocks arranged in three-stage subgroupings. The reed switch is a compact, two-wire, cardmounted matrix using a unique method of electronic operation and latching of the reed relays. Alternatively fully electronic switches are provided—of particular interest is the digital group selector which switches 8-bit PCM. The control uses a data processing system in a two-level processor hierarchy with one central processor and several regional processors. The number of regional processors is proportional to the size of the exchange. The central processor takes care of the complex functions, while the regional processors perform simpler, repetitive, functions such as scanning of test points. Fig. 1 AXE 10 row Fig. 2 shows the hardware structure of the AXE system and will serve as an illustration for a discussion of the system structure. The main subsystems of AXE are shown; it must however be understood that there are several additional subsystems which are designed in software only and thus are not shown in this diagram. Of the switching system, APT 210, three subsystems are shown and of the control system or, more correctly, the data processing system, APZ 210, also three subsystems. An important feature is that each subsystem forms an autonomous unit (hardware and software) of which the interfaces towards other subsystems are standardized. This means that subsystems may be combined in a number of different ways to match different applications, different functional mixes and to meet different traffic requirements. This concept of modularity is one of the most important characteristics of the AXE system and is found also at the lower levels of the system hierarchy. The subscriber subsystem, SSS, is arranged in modules for 2000 subscribers per group. Such a group (for 2000 subscribers or, by partial equipping, 250, 500 or 1000) may be located outside the main exchange and serve as a subscriber concentrator. The concentrator is connected to the parent exchange 55 over two or more 30-channel PCM systems w h i c h also provide the signalling facilities for the prolongation of the regional processor bus, RPB. The group switching subsystem, GSS, is designed with reed switches. The module in this case is one pair of RP's to control a group of 512 inlets and 512 outlets. While retaining the GSS subsystem interfaces the reed switchblocks may be replaced by a digital switch including in this case PCM terminals. For connection of digital lines digital interfaces are introduced. The trunk and signalling subsystem, TSS, can handle a very large variety of line- and register signalling schemes. First choice signalling systems in AXE are MFC (CCITT R2) and c o m m o n channel (of the CCITT No. 7 type), but virtually all other systems are handled, including SxS and different revertive pulsing schemes, so as to allow introduction of AXE with a minimum of adaptation in existing exchanges. Fig. 3 A X E 10 traffic characteristics Black curves show the traffic capacity for different SSN and GSN switch-block alternatives. Red curves show the call-handling capacity of the control system at 100 sees and 180 sees mean holding time respectively Fig. 2 A X E h a r d w a r e structure. 20,000-line exchange SSS LIC SSN AJC BJC RP RPB GSS GSN TSS ITC OTC ASD CRD CSD CPS CPU PS DS RS PT MAS MAU IOS TW CT Subscriber Switching Stage Subscriber Line Circuit Subscriber Switching Network A Junctor Circuit B Junctor Circuit Regional Processor Regional Processor Bus Group Switching Subsystem Group Switching Network Trunk and Signalling Subsystem Incoming Trunk Circuit Outgoing Trunk Circuit Auxiliary Services Device Code Receiver Code Sender Central Processing Subsystem Central Processor Unit Program Store Data Store Reference Store Processor Test Maintenance Subsystem Maintenance Unit Input Output Subsystem Typewriter Cartridge Tape Deck Traffic properties of the AXE system The traffic-carrying (erlang) capacity of the AXE system is demonstrated in fig. 3. The four horizontal lines of the diagram denote the four available SSS alternatives, for 0.09, 0.14, 0.20 and 0.30 erlang/subscriber respectively. The two black curves correspond to the traffic-carrying capacities of the two versions of the reed switch group selector; these two versions differ in maximum size only. For example, at a traffic of 0.12 erlang/subscriber the exchange is equipped with 0.14 erlang SSS groups and, with the larger GSS, may be extended to a maximum capacity of 50,000 subscribers. The second traffic parameter is the call handling capacity of the control system. This depends on a number of factors, in particular the functional mix and traffic distribution. For an exchange application with 100% key-set telephones, 5% intra-office traffic, in- 56 ter-office traffic using MFC signalling, certain volumes of abbreviated dialling, toll dialling, etc, the call handling capacity is 144,000 busy hour call attempts (BHCA). For an application with 0.10 eriang per subscriber and a mean holding time of 100 seconds per call this corresponds to a maximum exchange size of 40,000 subscribers. Fig. 3 includes two red curves to show the call handling capacity in relation to eriang capacity with functional mix as above and 100 and 180 seconds mean holding time respectively. For the very large applications a separate central processor is used for the outgoing group selector and associated trunk and signalling functions. With functional mix as above and a mean holding time of 100 seconds the maximum exchange size in this case is about 65,000 subscribers. Fig. 4 Magazine fully equipped with printed circuit boards for trunk circuits, control circuits and DC/DC converter is to say by commands from a teletype, or, for bulk-handling, from paper tape or magnetic tape. This capability provides the means for introducing remote control of the exchange or a number of exchanges from an administrative centre, Operation and Maintenance Centre, OMS. In such conditions SPC exchanges in a local network are not staffed. Changes in data by command, especially subscriber service categories and exchange data, replace the traditional strapping in conventional exchanges. Some further factors of interest when defining traffic and call handling capabilities of a given system are discussed in the "yellow box" on page 57. Other facilities, notably subscriber services, are available to cover a very large spectrum of requirements and may be introduced with the original installation or added at a later date. This is often done in the form of a program addition. In the "red box" on page 59. are listed examples of typical subscriber facilities. The "blue box" on page 59 lists as an example the available possibilities for adopting different types of charging schemes. Services AXE software system The characteristic property of stored program control is the relative ease in providing a very large range of subscriber services and administrative services. It is generally considered that the most important feature of an SPC system is the capability of introducing and changing different types of exchange data via electrical signals, that The important philosophy of the software of the AXE new generation SPC system is discussed in a separate article. The modular structure of the software provides ease of handling and software security. Basically the concept means that, with block modularity, there are no globally accessible data in the system through which software 57 Some factors affecting the traffic capacity of SPC switching systems The traffic capacity of the switchblocks and of the central processor naturally depends on the desired grade of service under normal load and under specified overload conditions. The maximum capacity of the system as given in this article is calculated with the following service quality: Fig. 6 Call-handling capacity, dependance on complexity and idle load tions with idle load above 20% are unlikely. The processing time per call depends on the functional mix of for instance, decadic and MFC signalling, the mix of dial and key-set telephones, the proportion of internal traffic, abbreviated dialling etc. The call-handling capacity quoted is valid for an environment of 100% key-set telephones, MFC signalling and 5% internal traffic. Higher capacities than 144,000 BHCA are obtained for cases requiring less processing power. The call-handling capacity of a medium complexity exchange is illustrated in fig. 6, showing a certain dependence on the volume of idle load. * Read: "Probability of delay exceeding 1 second". ' * Read: "Probability of loss". These service quality figures will comply with the majority of requirements. However, other requirements may be encountered, so the traffic capacity figures quoted should be looked upon as examples—more stringent service quality parameters will result in correspondingly reduced traffic-carrying capacity and vice versa. The call-handling capacity of the central processor depends on the idle load, the required processing time per call and on the grade of service (under normal load and specified overload conditions). The idle load which is a measure of the processor time not available for call handling, is effected by, for instance, various traffic recording and charging facilities. A normal value of the idle load is 10% and situa- The service criteria and overload definitions used in the calculations are as shown in the following table. Indicated in diagram by Solid lines Broken lines Probability of delay exceeding 1 second Normal Overload load 1.0% 0.1% 10% 1% The a and b lines in fig. 6 show the dependence on the definition of the overload: a) 20% increase of call intensity above normal value b) 40% ditto Normally the overload criteria determine the capacity. As a result the service quality under normal load is far better than required. As is evident from the diagram, the callhandling capacity for lower complexity requirements exceeds 200,000 BHCA. 58 faults may be dispersed via faulty data. Instead, the program of a given block can access data of the same block only. A possible software fault is immediately detected as an attempt to interwork by means of a faulty signal and so contained within the block. The block modularity has also been so arranged that the function block is the design and testing object of the system, thus increasing the handling capability—the programmer works only within one block at a time. Block interwork is microprogrammed, which eliminates the laborious address calculations and this inherent source of software faults found in earlier systems. AXE hardware reliability Fig. 5 Different kinds of printed circuit boards used in the switching system APT 210 The software security is matched by hardware reliability provided by a redundancy strategy of duplication and sectioning. Fig. 2 gives some indication of this scheme—the central processor is duplicated, with the two processors working in synchronous mode. The regional processors are duplicated; their mode of operation may be described as load-pairing, i.e. in case of failure in one RP the other one takes over the full load. In the switching system a sectioning scheme is used—this is designed so that the influence of a possible hardware fault will affect a maximum-size group of circuits only— such as maximum 16 trunks or 64 subscribers. Power is supplied through DC/DC converters; the provision of these follows the same hardware reliability structure —each CPU, each RP, each switching section has its own power source. Software support commands and printouts obeys the same syntax as the APS languages. This language is the Nordic MML, at present handled in CCITT as a joint contribution from the Scandinavian administrations. Naturally there are a great number of other aids for use in all phases of design, engineering, production, testing, installation and operation. Mechanical design The packaging structure, described in a separate article, mirrors the modular concept of the AXE system. This means that the functional module of the system—in this case the function block hardware—corresponds to the mechanical module, the magazine. In order to allow for the varying hardware content there are different sizes of magazines. This packaging system structure provides the important feature that the equipped magazine—as it constitutes a well defined functional entity, the function block—is functionally tested together with its proper software before leaving the production phase. As each magazine is shipped fully equipped, it will arrive on the installation site as a fully tested unit; installation time has been greatly reduced. Cost effectiveness Cost effectiveness in telecommunications equipment means minimization of total cost. In general terms this is to say that the sum of the yearly costs for equipment, installation and operation shall be as low as possible. The software support is made up of a number of aid systems which form part of AXE. The most important is the AXE language family including the highlevel programming language, PLEX. PLEX is defined within the programming system, APS, which is the main software support for AXE software processing. This has been the leading philosophy for the development of the AXE system and one consequence has been a very determined focusing on handling. Efficient handling is ensured in different ways, chiefly by creating a modular system structure and by developing powerful aids. This system philosophy is important in all parts of the system and many references will be found in The man-machine language of AXE for the fnllnwinn arHr-lpc 59 Subscriber facilities in AXE The following list of available subscriber services in AXE is not complete—it is included to exemplify the very wide range of functions in the system. —• Call barring (restricted service), outgoing traffic. Subscriber- or operator-controlled. — Call barring, incoming traffic. — Absent subscriber service. Rerouting to announcing machines or operator. Subscriber- or operator-controlled. — Interception service. Rerouting to announcing machines. — Pushbutton dialling. Dial and pushbutton sets in parallel on same line if desired. — Malicious call tracing. Automatic recording or controlled by last party release with manual tracing. Solution depends on network capabilities. — Abbreviated dialling. Operatoror subscriber-controlled. — Hot line service — Subscriber's private meter — Priority — Hold for enquiry and call transfer — Add- on conference — Call waiting — Automatic alarm call service. Alarm clock. — Information services. Recorded announcements. — Direct in-dialling to PABX's —• Coin boxes — Centralized PABX (Centrex) Call charging functions in AXE The AXE system will accommodate any desired charging scheme: flat rate, one metering pulse per call, multimetering, toll ticketing or a combination of two or more. Charging may be performed on local calls only (in which case toll calls are charged higher up in the network hierarchy) or on local calls plus all or parts of the national traffic. Charging methods — Pulse metering; different schemes, for instance according to the Karlsson principle. — For pulse metering the output is produced, on command, on magnetic tape or data link to central processing. Any individual meter, or a group, may be read out by using special commands. - Toll ticketing — LAMA (Local Automatic Message Accounting). — Continuous output per call on magnetic tape or data link to central processing. Charging facilities — Call specification on request from subscriber (immediate service), also for pulse metering. — Recording of outgoing calls for charging control. — Charging to B-party (collect). Obtainable with toll ticketing. — Subscribers' private meters. Introduction of the AXE 10 Switching System in the Telephone Network Staffan Braugenhardt, Inge Jonsson and Bo Nilsson In the majority of applications a new exchange system is introduced into the environment of an existing network of exchanges and an existing administrative system. But it must also be designed with a view to the new concepts of tomorrow. How the AXE 10 switching system fulfils these objectives is discussed below. The main part of the telephone network of today is analogue and will remain so for many years. For networks based on two-wire transmission, the AXE system offers analogue two-wire switching with high quality transmission performance. UDC 621.395.722: 681.3.065 621.395.7'! LME 83022 8305 Survey When introducing a new telephone exchange system into the telephone network, it is necessary to make sure that the system will meet the demands of today and also of tomorrow. The system must be designed for a long and useful life, fig. 1. Fig. 1 AXE 10 is well adapted for operation in an environment with increasing facility level Fig. 2 AXE 10 is an efficient and versatile component both in the analogue network of today and in the digital network of tomorrow used in an economic way both for exchanges with normal subscribers and for exchanges with advanced business subscribers. The AXE system can be introduced into an existing telephone network as a self-contained switching point without affecting the surrounding exchanges or equipment. But the AXE system also meets far-ranging requirements for future networks. It is well adapted for convenient transition of the networks of today into the future digital networks with advanced subscriber services and centralized, computer-aided operation. Today most administrations primarily supply the ordinary telephone services. But facilities such as push-button dialling, abbreviated dialling, in-dialling to PABXs and extended interception are becoming increasingly attractive and are likely to be standard in the future. In addition there is a whole range of services of less obvious general importance but of great interest for selected subscribers. One such group of facilities, intended primarily for public administrations and the business community, includes for example centrex, call transfer, automatic call back and follow me. The philosophy for the provision of subscriber services in AXE exchanges is a basic system with normal facilities and optional add-on facilities which do not affect the cost of the basic system. This means that the AXE system can be In the long term, however, digitalization of the network will take place, with PCM transmission and digital switching. The digitalization will result in a new network structure. This structure can be described as a hierarchy of digital group selectors at a small number of levels to which subscriber stages—concentrators—are connected. The concentrators are decentralized and placed near the subscribers. Since the AXE system includes both analogue and digital group selectors, and has subscriber stages which can be placed either close to or away from the group selectors, the AXE system is designed to assist in the transformation of the analogue networks of today into the future digital networks, fig. 2. Telephone companies are becoming increasingly computerized in their administration. New exchanges must fit into this new automated environment. They must be able to communicatewith the administrative system; that is, they must be able to supply data to and receive and execute orders in cooperation with the various parts of the administrative system, such as computers, terminals and staff. The administrative systems now beginning to take shape envisage not only functions for sophisticated maintenance but also considerably increased facilities for fully automated charging and advanced traffic recording. In this respect, too, the AXE system is well adapted for integration in a total administrative system. The software structure and the input/output organization of the AXE system are designed for the future possibility of communication with network operation centres and subscriber service centres via data links. STAFFAN BRAUGENHARDT INGE JONSSON BO NILSSON Telephone Exchange Division Telefonaktiebolaget LM Ericsson The role of AXE in the continuous modernization of city telephone networks The AXE system can be introduced into an existing metropolitan network as a self-contained switching point without affecting the surrounding exchanges and equipments or requiring changes oradditions(except normal extensions) in other parts of the network. However, the AXE system also offers a number of facilities which make it possible to simplify and improve the network as a whole. Some of these improvements concern more efficient traffic handling functions (signalling, routing etc.), others are related to charging, operation and maintenance features. Fig. 3 shows the introduction of AXE 10 in an existing multiexchange area. A number of typical AXE local exchange characteristics of importance when introducing the system in existing telephone networks maybe mentioned: Fig. 3 Simplified illustration of the introduction of an AXE 10 exchange in an existing multiexchange area, applying prevailing signalling principles for interwork with existing exchanges XB SXS MFC Crossbar exchange Step-by-step exchange and signalling MFC signalling Naturally two-wire switching is provided, with high quality analogue transmission performance between inputs and outputs of the exchange. This feature guarantees that traditional transmission planning rules and objectives used for the unamplified part of the national network today can be maintained. As mentioned earlier, AXE also comprises digital switching facilites for proper use in the increasing digital environment of tomorrow. Subscriber-loop interfacing facilities are included, with wide range for signalling and power feeding, making it possible to cooperate with the large variety of existing analogue telephone instruments. Special attention has been paid in the AXE system design to the use of push-button dialling in the subscriber network and also to the expected situations where a mixture of pushbutton and rotary dial instruments are connected to the same exchange. AXE permits the inclusion of a large number of subscriber facilities by the addition of functional blocks to the basic system. This has proven to be an economical way of solving the great variety of different needs in different national networks. Examples of modern facilities obtainable as optional add-on functions in AXE are automatic alarmcall service, abbreviated dialling and centrex. AXE is equally suitable for operation in Multi-Frequency Code (MFC) networks and in the coming Common Channel Signalling (CCS) networks. The modular build-up of the system facilitates its adaptation to any existing or planned signalling method, e.g. older type AGF network revertive pulsing or special national systems such as the Socotel multi-frequency signalling. Furthermore the AXE system is easily adapted to any existing type of trunk and junction interface for speech and line signalling. An increased number of routing possibilities become available when AXE is introduced. Conventional exchanges are limited in this respect due to a number of factors, such as B-number analysis depth, number of routes, number of circuits per route, number of rerouting possibilities, etc. A modern SPC exchange, such as AXE, has practically no limitations of these kinds. Other new possibilities available in AXE are the flexible traffic control features. These make it possible to change the routing of the exchange by command, either locally or from a central point in the network, so as to adapt to traffic load changes and possible disturbances. The charging facilities are easily adapted to the procedures already used in the network, e. g, with a charging point for multimetering or toll ticketing situated in a superior centre. All facilities required for charging of calls and subscriberservices may, however, be included in the AXE local exchange itself. This means that charging analysis, determination of tariff, periodic pulse metering, call specification and toll ticketing can be performed by AXE. AXE is easily adapted to the existing operation and maintenance organization of an administration. A single AXE installation in an existing network may 62 be regarded as a self-contained item with respect to operation and maintenance, i.e. all required features of this type have been included in the basic system. It is however foreseen that the system will be serving to a great extent in networks with centralized operation and maintenance. It is, of course, not possible to utilise all the new facilities at the introduction of the first single AXE installation in an existing network unless the cooperating exchanges have similar capabilities. The concept of updating existing crossbar installations to SPC standards has been discussed in Ericsson Review No. 4 1973' and 3/4 19752. AXE is well adapted to such a modernized environment, at the same time forming an attractive link to future network concepts. AXE as base for future digital networks The digital trend Digital transmission in the telephone network is by now well established. Partly as a consequence of digital transmission and partly as a competitive technology in itself, digital switching will find an increasing number of applications. The combination of the two will result in integrated digital networks. Fig. 4 Comparison of typical total cost curves for transmission and switching techniques ^ ^ B B ^ Economical range for PCM with analogue switching 4^^^m^ Economical range for PCM with digital switching The effect of this integration is schematically demonstrated in fig. 4. The figure shows the cost per circuit for different transmission media combined with analogue and digital switching. Assuming for the sake of discussion that a digital group selector (including A/D conversion) and an analogue group selector have the same cost in an analogue environment, the curves for voice frequency (VF) and frequency division multiplex (FDM) will be independent of the switching technology. For the case with analogue switching the interceptions of the two lines representing VF and FDM with the blue PCM line will mark the economical range of PCM (blue arrow). This will normally affect only small portions of the junction network. The combination of digital switching and PCM transmission gives the area represented by the red arrow. Obviously the economical range for PCM is increased. We see that PCM becomes competitive compared to VF over all distances and compared to FDM over a much wider area than in the case of analogue switching. Clearly the introduction of digital switching supports the use of digital transmission. The use of digital transmission is, however, one cf the major reasons for the introduction of digital switching. It is this interactive effect which will lead to the growth of integrated digital networks. 63 Development towards a digital network Digital switching and transmission equipment can be introduced into the telephone network in three main ways. It can be overlaid over existing equipment, it can replace existing equipment, and it can be used to extend the network into new areas. In principle the overlay method means that the analogue network is extended with digital equipment in such a way that an integrated digital overlay network is formed. For practical applications it seems probable that none of the methods will be used exclusively; rather it is expected that combinations of the methods will be used in different parts and development stages. An example of development towards a digital network is shown in fig. 5, where a) The starting point is an analogue network with two local and one tandem exchange. Analogue local exchange Analogue tandem exchange AXE 10 local exchange with digital group selector AXE 10 digital tandem exchange PCM terminal VF transmission PCM transmission Subscribers' stage or concentrator Fig. 5 D e v e l o p m e n t t o w a r d s a digital network ( T h e c o n c e n t r a t o r s are assumed to be of a n a l o g u e type) Fig. 6 A X E 10 w i t h digital g r o u p s e l e c t o r and r e m o t e s u b s c r i b e r stage SSS-C CLI LIC SSN GSS-D GSN-D PCD CPS BP ST SSS TSS ETC |TC OTC CSD CRD Subscriber Switching System-Concentrator Concentrator Line Interface Subscriber Line Circuit Subscriber Switching Network Group Switching Subsystem, Digital Group Switching Network, Digital PCM Terminal Central Processing System Regional Processor Signalling Terminal Subscriber Switching Subsystem Trunk and Signalling Subsystem Exchange Terminal Circuit Incoming Trunk Circuit Outgoing Trunk Circuit Code Sender Code Receiver b) Extension of the transmission circuits is done with PCM. The digital signals are converted to analogue before they are switched. c) Further extensions of the transmission circuits are made with PCM links. The increased digital environment makes it feasible to introduce two digital switching centres, one local exchange in a new area and one tandem exchange as an extension of the existing one. We here get the first complete digital link, without A/D conversion between two exchanges. d) One of the analogue exchanges is replaced by a digital exchange. A remote subscriber stage, a concentrator, is introduced. For stage c and onwards the network synchronization problems must be solved. This can be done by plesiochronous working, i.e. with autonomous clocks in the exchanges, by appointing one of the exchanges a master or by using mutual synchronization. System AXE has been designed to serve as a basis for digitalization of networks. As an alternative to the reed switch the system may be equipped with a digital group selector, thus giving the administration the choice of the most economic network solution, based on actual and planned proportion of digital versus analogue equipment in the network. A block diagram showing AXE with digital group selector and remote subscriber stage is shown in fig. 6. A fully digital network In the long term it is expected that the digitalization will penetrate down to the subscriber network including the subscriber set. This has potential merits both in the economic and technical spheres. Economic gains are obtained, as expensive technology interfaces in the subscriber stage can be eliminated. Technical gains will be noise and crosstalk elimination, four-wire working on the single pair, transmission level consistency and data capability. 64 Signalling in AXE networks General remarks on signalling in AXE Signal interchange with the environment is performed by two subsystems of AXE. The trunk and signalling subsystem (TSS) is responsible for signalling on trunks and junctions. The subscriber switching subsystem (SSS) performs the signalling funtions required to and from subscriber equipment, fig. 7. Adaptations can be made to any signalling method by the design of proper hardware and regional and central soft- Fig. 7 S i g n a l l i n g f u n c t i o n s in an AXE e x c h a n g e a d a p t e d to a m i x e d MFC a n d step-by-step network a b Receipt of digits from A-subscriber, setting up to wanted route and signalling over j u n c t i o n line Exchange of signals over i n c o m i n g j u n c t i o n line and setting up of SSN to the B-subscriber In neither case is the establishment of the c o n n e c t i o n s through GSS shown. SSS GSS TSS LIC AJC KRD CSD OTCi OTC2 BJC CRD ITCi ITC2 Subscriber Switching Subsystem Group Switching Subsystem Trunk and S i g n a l l i n g Subsystem Off-hook detection Detection of decadic pulses etc. Detection of tones from push-button instruments Code Sender for MFC s i g n a l l i n g on o u t g o i n g trunks Line s i g n a l l i n g on outgoing trunks w i t h register s i g n a l l i n g of MFC type Step-by-step s i g n a l l i n g on outgoing trunks Injection of ringing signal and tone, off-hook detection etc. Code Receiver for MFC s i g n a l l i n g on i n c o m i n g trunks Line s i g n a l l i n g on i n c o m i n g trunks w i t h register s i g n a l l i n g of MFC type Step-by-step s i g n a l l i n g on incoming trunks ware within these subsystems without affecting other parts of the total system. Most variations will, of course, apply to the signalling requirements to and from other exchanges in existing networks. A number of different function blocks (MFC signalling with the associated line signalling method, AGF revertive pulsing, decadic pulsing for step-by-step interwork, etc.) are therefore available within TSS in various combinations. The functional subdivision of the AXE system is extremely practical for engineering and design work, e.g. for Fig. 8 C o m m o n c h a n n e l s i g n a l l i n g C C S i n AXE 10 tandem exchange GSN-D ETC PCD ST RP G r o u p S w i t c h i n g Network, Digital Exchange Terminal Circuit PCM Terminal S i g n a l l i n g Terminal Regional Processor 1. 2.4 or 4.8 kbit/s 2. 64 kbit/s 3. T h r o u g h - c o n n e c t e d s i g n a l l i n g time-slot Speech path 2.4 or 4.8 kbit/s s i g n a l l i n g path 64 kbit/s s i g n a l l i n g path projects with new signalling requirements. Some aspects on the introduction of AXE in an existing MFC-network The MFC network is a concept comprising not only the MFC signalling system itself but also the wide range of functions and facilities obtainable in a network with MFC signalling. The features of the MFC technique, such as flexible call set-up procedure in networks with unknown number length, transfer of information on Asubscriber category and B-subscriber line condition, etc. are well known. In many national networks all possibilities inherent in the MFC system are not utilized, partly because they are not needed and partly because existing installations are not capable of treating all signals that could be transmitted by MFC. When SPC technique in the form of AXE is introduced in an existing MFC network, a number of new possibilities, obtainable with MFC but not utilized in existing plant can be considered. Among the facilities made available by the combination of SPC and MFC are: — extended use of A-categories — extended use of B-conditions — use of signals specifying the cause of congestion etc. Some fundamental MFC signalling procedures in existing national networks can be simplified or eliminated when AXE is fully utilized. There are, for example, extra MFC sequences required for charging purposes because the charging point is located higher up in the network. Backward MFC signals (e.g. B-subscriber free, no charge) as well as forward MFC signals informing about the A-subscriber identity (for toll ticketing) must be monitored or received in the charging point. These additional functions during call set-up are today normally solved by having a regenerating and controlling MFC register in the charging point remaining connected during the whole setting-up process. AXE may gradually eliminate complications of the above type, as it is capable of performing all required charging functions itself and allows the charging point to be located in the local exchange. The total call set-up time and the postdialling delay in an MFC network are directly dependent on the time requiered for the connection of MFC equipment and the through-connection time of the switching stages. This is a consequence of the end-to-end principle and of the compelled procedure in MFC signalling. AXE here offers further improvements in existing MFC networks as the above mentioned switching delays are very short compared to those in conventional exchanges. Common channel signalling (CCS) in AXE When exchanges with stored program control are introduced on a large scale, the most attractive signalling system between them is common channel signalling (CCS). A new CCS system primarily intended for use in digital networks is presently being developed (CCITT No. 7). Although the system will be specified for international circuits, it is foreseen that the main applications will be in national networks. An international specification is expected to be completed in 1979—80. A national application of this CCS system will be used as standard in many future AXE networks. The principles developed for the new CCS system are also applied in the signalling between the AXE parent exchange and subscriber concentrators. The No. 7 CCS system will be optimized for use with the signalling rate of 64 kbit/s employed on PCM links. The high bit rate permits simple procedures for error control. Although the signalling system is optimized for 64 kbit/s, it is also suitable for much lower rates (e.g. 4.8, 2.4 kbit/s) derived from analogue circuits equipped with modems, fig. 8. When these lower rates are used, the maximum number of circuits served by the CCS link will be limited, but still sufficient to handle large routes. The new CCS system will thus be generally applicable between AXE exchanges irrespective of the type of transmission medium (digital or analogue) used in the network. 66 channel signalling network is built up, interconnecting the new exchanges. In this CCS network each route may have a regular and a stand-by link of its own (associated signalling) or some routes can be served by signalling links connected via another exchange (non-associated signalling). Advantages of Common Channel Signalling There are a number of reasons why CCS is expected to become a standard signalling method in future AXE networks. As mentioned earlier, the SPC nature of AXE makes it possible fully to utilize the information capability of the MFC signalling system. However, Common Channel Signalling together with SPC is an important step towards the introduction of more advanced functions and facilities in the network. The high signalling capacity of CCS systems means that a greater variety of Acategories, B-subscriber line conditions etc. can be transferred than in the MFC case. Moreover, total call set-up time and post-dialling delay are reduced as compared to MFC. The performance offered by MFC in this respect is, however, sufficient for telephony. A CCS link intended for telephony signalling can also carry information for other purposes. Additional services involving communication to specialized centres can easily be introduced into the network thanks to the flexible system structure of AXE and the presence of CCS facilities. CCS signalling between AXE exchanges is expected to be more economical than other signalling methods. I n s t a l l a t i o n of t h r e e AXE 10 e x c h a n g e s in an existing metropolitan network, thereby i n t r o d u c i n g the first s t a g e of c o m m o n c h a n n e l s i g n a l l i n g , CCS This figure illustrates the performance of all translation between CCS and MFC in exchange A. For security and other reasons this may, however, not be suitable in large metropolitan areas. Signals between A and B are normally transmitted over path A—B (associated signalling) while path A—C—B may be used as stand-by (non-associated signalling). a b c Existing crossbar exchanges Installation of the first AXE 10 exchange retaining the existing MFC s i g n a l l i n g system, Installation of another two AXE 10 exchanges and i n t r o d u c t i o n of CCS between all AXE exchanges CCS links - ^ — speech paths Introduction of Common Channel Signalling together with AXE in existing networks Although CCS is a very attractive future signalling system for AXE, complications would be experienced if CCS links were to be connected to conventional exchanges. SPC is more or less a requirement for CCS signalling. It may therefore be questionable whether or not to use CCS at the introduction of AXE in an existing network, e.g. an MFC network. Some basic rules facilitating the use of CCS at an early stage may be outlined. See also fig, 9. • Signalling between all AXE exchanges introduced into the existing network is by CCS. A common • The number of translation points between existing signalling and CCS is minimized, so that a limited number of the new exchanges need be equipped for interwork with the existing network. Calls between the new and the existing parts of the network are routed via these translation points. The basic rules outlined above and the illustration in fig. 9 are not intended as the only solution of the mixed CCS and MFC network. They indicate merely that there are reasonable ways open to an administration which prefers the use of CCS at an early stage when AXE is introduced into a network. It should be noted that the existing network used in the illustration could be an MFC network, but also any other type of network such as AGF or Stepby-step. Subscriber services A few special subscriber services are included as standard in the basic version of AXE. One example is a simple form of interception, without operator's assistance, but with rerouting to a recorded announcement or tone information. A large number of future services are offered as options, which can easily be introduced due to the modular system structure of AXE. Most of the subscriber services in modern and future networks are performed by the AXE local exchange itself, others will involve communication procedures using other parts of the network. Services of the first type are abbreviated dialling, automatic alarm-call service, hold for enquiry, etc., as well as PABX services including centrex. Absent-subscriber and interception services with operator assistance in a SDer.iali7ed centm are 67 F i g . 11 S e q u e n c e for subscriber-controlled insertion of a new a b b r e v i a t e d number * Prefix for access to subscriber service XX 2-digit service code 7 New abbreviated number to be inserted 08719 0 0 0 0 examples of the second type. In these cases AXE will be equipped and prog r a m m e d with all functions required for rerouting to and c o m m u n i c a t i o n with such a centre. The communication procedures used between the subscriber and the AXE local exchange for these services are in accordance with the standard procedures proposed by CEPT. Most of the advanced new services require telephone instruments with a 12-button keyset. Connection of such instruments is a standard facility in AXE. Corresponding new B-number a Suffix indicating end of message Services such as abbreviated dialling, automatic alarm-call and absent subscriber are becoming more and more important for administrations and telephone users. These services and the method of their performance in AXE are described in greater detail below. Abbreviated dialling Abbreviated dialling means simplified dialling for connections to frequently used telephone numbers, by replacing these numbers with one- or two-digit abbreviated numbers. The abbreviated numbers are translated into full local, national or international numbers in the originating exchange, fig. 10. A subscriber can have a maximum of 100 abbreviated numbers, w h i c h may be either one-or two- Fig. 10 A b b r e v i a t e d dialling in A X E 10 1 The abbreviated B-number is analysed and translated into an ordinary B-number by the central processor CP 2. The ordinary B-number is used during the setting up in the network digit. By use of the so-called suffix method, mixed one-and two-digit n u m bers may be e m p l o y e d . Alterations in the abbreviated number list are made by the a d m i n i s t r a t i o n . As an optional feature such alterations can also be made by the subscriber, who will then proceed a c c o r d i n g to fig. 11. Automatic alarm — call service In this service a call is automatically generated and a wake-up announcement issued to a subscriber having requested this max. 24 hours in a d vance. This facility is accessible to all subscribers and can be booked either directly from his own telephone set or via an operator. The subscriber can also cancel the bookings himself for the next 24-hour period. The subscriber is rung either for 1 minute or until the time supervision expires. On encountering busy or no reply, three call attempts are made during three consecutive 5-minute intervals. If all attempts are unsuccessful, a notification to that effect is automatically written on a typewriter. Absent-subscriber service At a subscriber's request, calls made to his number during a specified 68 period can be intercepted and diverted to a special equipment or to an operator. Rerouting facilities for this purpose are standard in AXE. The special equipment (special information tone generator or announcing machine) can be located in the exchange, while the absent-subscriber operator is normally stationed in a specialized centre covering a number of exchanges. The normal procedure for requesting the facility is that the subscriber contacts the administration. He states his subscriber number, the time and durtion of the desired service and the form it is to take. The exchange is " p r o g r a m m e d " accordingly by the staff. F i g . 12 Operation and Maintenance Centre OMC with a s s o c i a t e d f u n c t i o n s OMC Operation and Maintenance Centre. For operation and maintenance activities of the exchanges in the network The functions below may either be included in OMC or form Independent centres. SA SC + FD SS CH TM TA NM Service Assessment. For evaluation of service quality provided to subscribers Subscriber Complaint and Fault Dispatch. Responsible for the subscriber network Subscriber Service. For provision and administration of special subscriber services Charging service. For c o l l e c t i o n and processing of charging information Transmission Maintenance Responsible for the trunk network Traffic Analysis. For c o l l e c t i o n of statistical information for planning and for analysis of the network performance indicators Network Management As an optional add-on facility the service can be made subscriber-controlled. This means that the subscriber can connect to and disconnect from the service by dialling special codes. Centralized operation and maintenance To cope with the rapidly growing size and complexity of telephone networks, combined with rising costs for personnel and equipment, an expanded use of automated equipment becomes necessary. A consequence of this development is the use of computers for the control of switching systems and for administrative routines, which in turn leads to the logical step towards network operations centres (NOC). Certain aspects of the SPC systems make this step technically and economically feasible. Operational activities such as observation of system performance and changes of system parameters relate mainly to software and may consequently be performed remotely. This means that staff must be present in the exchange only when hardware maintenance, such as printed board replacement and work on the MDF, is needed. All data transmission to and from the exchange, including internal and external alarms, is carried over one reliable, medium-speed data link. The reliability and automated mainte- nance characteristics of SPC exchanges give the personnel of a single exchange very limited practice in troble analysis and fault localization. To keep up the proficiency of the maintenance personnel it is advisable to centralize them in operation and maintenance centres. Job satisfaction is gained by the removal of routine and paper work. Tricky and boring analysis of paper outputs is eliminated. Centralization also makes it possible to collect information not only about exchange performance but about network performance as a whole, which provides new possibilities for supervision, planning and d i m e n s i o n i n g . A major advantage of centralization is that activities can be coordinated between sales, operations and planning departments. The traditional barrier between transmission and switching people is lowered. They can be colocated in operation centres or they can work with identical information from the network or from data bases. Through this c o o r d i n a t i o n , and making use of computer aids in central positions, the administration will be able to attain better service at lower cost, with all activities based on up-to-date network information. Administrative networks As described above, the AXE 10 system can be introduced in an existing network as a self-contained switching point. This is, of course, also true of operation and maintenance, in which resoects AXE 10 offers all facilities required by a modern SPC exchange. It is also possible to integrate the system as a building block in administrative networks. How this will happen depends on several factors such as network hierarchy, degree of automatization of switching functions, use of SPC exchanges and existing organization. Fig. 12 shows the connection of AXE 10 to an operation and maintenance centre OMC. Different administrative cent r e s s u c h as Central Sales Office, Billing Centre and Directory Enquiry Centre 69 Fig. 13 Configuration of the O M C e q u i p m e n t in a large AXE 10 network The alarm display (T) shows the alarm situation in the whole area. From the operator's position © bothway communication Is established with the individual exchanges and detailed information is received about different alarms etc. Printers can also be connected to the information processor. The communication processor transmits the bothway data between the telephone exchanges and the information processor and thus relieves the load on the latter. The information processor handles all data. Both processors are interconnected by busses. Operation and Maintenance Centre, OMC Fig. 12 also shows the functions associated with OMC and fjg. 13 the configuration of the OMC equipment in a large AXE 10 network. The different functions shown in fig. 12 may either be included in OMC or form independent centres. OMC is a control point for several exchanges and also handles the data communication with the administrative data system and with other specialized network operation centres. The OMC equipment must be able to interwork with all exchange systems in the area and will: — interface and automatically switch all messages between operators, exchanges and data bases — perform language conversion to command languages appropriate to the different categories of operators — log all I/O activities that are important as background information for maintenance and updating of data bases — adapt all outputs to formats that suit each administration — automatize various functions resulting from operating experience, e.g. a fault library which is consulted automatically in typical fault situations — store an operator's guide that is used to simplify the work of the operators. The computers used in OMC are general purpose minicomputers which can be programmed in high-level language for appropriate functions. In large networks it is advisable, for capacity resons, to divide the functions between different computers. Routine and standard functions such as data link control can be handled by a communication processor. Data base- and manmachine-oriented functions are handled by an information processor. The alarms are displayed on a colour TV monitor which gives a survey of the alarm condition in the network, with the highest alarm state displayed in colourcoded form. Paper work is kept to a minimum through the use of automatic logging, analysis and search functions in the information processor. Paper outputs are received automatically, or on command, only when necessary. In an initial stage the functions shown in fig. 12, can be performed from terminals connected to the OMC computer system. The main benefits of centralization, as presented earlier, will be available at this stage. In an extended stage with independent centre equipment, such as TACS (Traffic Analysis Centre System), SCCS etc. it is possible to fulfil the special needs of all categories of personnel even in large networks. AXE 10 - System Description Mats Eklund, Carl Goran Larson and Kjell Sorme This paper surveys the background and requirements of AXE 10 and presents the resulting architecture. It then describes the two main systems that make up AXE 10, the APT 210 switching system and the APZ 210 data processing system. UDC 621 395 722: 681.3.065 LME 83022 I. SYSTEM SURVEY Functional requirements Stored-program-control switching systems have existed for a number of years and valuable experience has been gained both by operating administrations and by manufacturers. Careful analysis of this experience has provided the basic requirements for the AXE system. Minimization of total system cost, i.e. including installation and operation, must take precedence over hardware minimization. Cost trends today show that hardware costs, in particular IC components, will continue to decrease, while all types of labour costs are expected to continue to rise. This pattern has led to a systematic effort to improve the manageability of the system —to the creation of a system with favourable handling properties. Handling is a general term which includes all phases—design, engineering, production, installation and operation. Telecommunications today are characterized by dynamic progress both in technology and in the exchange environment—the networks. Thus the system requires functional modularity, so as to be introduced easily in many different and changing environments, and technological modularity, which permits the use of improved techniques within the system framework. Dependable operation is expected of any modern switching system. It was recognized that the reliability requirements must extend to the software as well as the hardware—thus the reliability strategy of hardware redundance is complemented by a corresponding requirement for software security. The final basic demand was for modular capacity, so that the system may be adapted to a very wide range of sizes, traffic conditions and functional mixes. The maximum capacity specified for a typical local exchange application is 40,000 subscribers at 0.10 erlang/subscriber and 100 seconds mean holding time. The fundamental requirements formulated above have had a decisive influence on the design of the AXE system. The following section presents the resulting principles of the characteristic AXE system structure and its composition and features. Functional structure The functional structure of the AXE 10 system forms a four-level hierarchy: system, subsystem, function block and function unit, fig. 1. The characteristic feature of the functional structure is the integration of both hardware and software in the functional modules; these are complete as regards their functional role in the AXE system. Each module may be seen as a "black box" in the system hierarchy, having standardized interfaces with the other modules, or boxes, on the same hierarchy level. This means that the module is known by its interfaces and it is not necessary to know its division into subordinate modules. At the function block level, for instance, system knowledge is provided by the block interface descriptions; the function units are in this context not of interest. At the top functional level AXE 10 is made up of the switching system APT 210 and the data processing system APZ 210. AD-r 0 i n •_ ^ T 2 1 0 W l t n Subsystems The switching system APT 210 is divided into six subsystems, fig. 2. Each subsystem is made up of hardware and software or of software only. The hardMvoro nm«iHog t P o o i Q ^ t r ; ^ i intorfar.ps A MATS EKLUND CARL GORAN LARSON KJELL SORME ELLEMTEL to subscriber and trunk lines and certain simple functions. The software is divided into regional software for simple and high-frequent functions and central software, which performs the more complex functions of lower frequency. The division between regional and central software is discussed in a later section. Each subsystem comprises a group of related functions. The subsystem grouping has also taken into account the desire to minimize the number of subsystems which are affected by differing environmental requirements— for example subscriber line specifications, signalling interwork with other exchanges, and man-machine input/ output. Adaptation to different conditions affects chiefly the subsystem SSS and TSS in fig. 2 as well as IOS in fig. 3. Technological developments or alternatives, such as the digital group Fig. 2 Subsystems of switching system APT 210 Subscriber Switching System Group Switching Subsystem Trunk and Signalling Subsystem Traffic Control Subsystem Charging Subsystem Operation and Maintenance Subsystem selector may in a similar way be introduced in a subsystem without affecting any other subsystem. Each subsystem is made up of a number of function blocks, each of which in turn consists of a number of function units. At the lowest level, the function unit, the rule is that each unit is implemented in either hardware or in software, while at the three higher levels hardware and software are integrated. The functional hierarchical structure of strictly defined modules may be seen as a "box of bricks". Engineering of the individual exchange is greatly facilitated as everything necessary to realize a given function (or group of functions) is contained within the module and alternative subsystems or function blocks may be combined in a great number of ways to meet each individual specification. 72 APZ 210 with subsystems The data processing system APZ 210 is structured on the same principles as APT 210 and consists of four subsystems, fig. 3. The central software of the APT switching system is executed by the central processing subsystem, CPS. CPS is made up of a central processor and the operative software of, for example, the monitor, loading and program test functions. The regional processing subsystem, RPS, executes the regional software of APT and is similarly made up of the regional processors and their operative programs. The maintenance subsystem for APZ 210 is a self-contained subsystem, MAS, containing functions for fault analysis, fault correction and diagnosis. The fourth subsystem of APZ 210 is the input-output system, IOS, containing functions for interworking between the AXE 10 system and operational personnel or other data processing systems. Input-output is performed either via alphanumerical or file-oriented IO devices. Controlled APT hardware Fig. 3 Subsystems of data processing system APZ 210 RPS CPS IOS MAS Regional Processor Subsystem Central Processor Subsystem Input/Output Subsystem Maintenance Subsystem 73 The diagram in fig. 4 shows the relationship in a telephone switching system between frequency of functions and their corresponding complexity. This analysis explains the background to the two-level processor hierarchy. The requirements for a favourable cost modularity coupled with high capacity is met by a structure with two types of processors for different tasks. The central processor is designed for handling complex functions and the regional processor is designed to handle simple, repetitive, high-frequent tasks. The number of regional processors depends on the size of the exchange. Frequency Fig. 4 Relationship between frequency of functions and their complexity The function block is the most important handling module and usually the subject for new design. The data processing system APZ 210 is structurally and functionally designed to match the function block concept. The software of a function block is consequently produced, loaded and executed on a block basis. The hardware structure has also been adapted to the functional structure, and the AXE packaging structure (BYB 101) conforms to the modular properties of the system. Hardware structure The structure of the AXE hardware is shown in fig. 5. It is characterised by a hierarchic division into switching hardware, regional processors and a central processor. Extension modules Fig. 5 The structure of AXE 10 hardware with duplication down to extension modules The three types of hardware interwork via bus systems. There is a bus system between the central and the regional processors, to which all regional processors are connected, and an EM bus (Extension Module bus) between each regional processor and the corresponding switching hardware. Fig. 6 Software interworking between function blocks The reliability structure of the AXE system, see fig. 5, is based on duplication all the way down to the extension modules. This ensures high reliability, a single fault affecting only one extension module. The central processor works in a synchronous parallel mode, whereas the regional processor works in a "pairing" mode; if a fault occurs the other processor of the pair takes over the full load. The bus systems are duplicated, both between the central processor and the regional processors and between the regional processors and the extension modules. The power supply system is sectionalized, which means that DC/DC converters are included in each extension module, and that each processor has its own converters. Software structure The software structure in the AXE system follows the functional structure, in which each module can be considered Fig. 7 Testing of central processing units in the development laboratory as a "black box . The data processing system APZ 210 has been designed on the basis that this functional structure is retained in the software down to the machine code level during normal operation. The central software parr of a function block consists of programs with associated data, which are stored in the central processor stores. The programs belonging to a block have access to data in the same block only. Interwork between different blocks is realized by transfer of messages between the programs in these blocks, so-called signals, fig. 6. The structure of the regional software is based on the same principles as the central software, but the implementation in the regional processor is simpler, as it handles simpler and smaller programs and less data. A new high-level language, PLEX, has been developed for AXE in order to simplify programming and the understanding of the programs. The processors in AXE have been adapted for PLEX in order to achieve efficient program handling while maintaining high system capacity. The high-level program language also contributes to high program reliability as it makes possible high quality software for the AXE system. 75 II. SWITCHING SYSTEM APT 210 The switching system APT 210 performs the AXE 10 system traffic functions and associated functions for operation and maintenance. The APT 210 structure is the same whether the application is a local exchange, a tandem exchange or a transit exchange. The present description is based on the local exchange application. Subsystem structure As already noted and as shown in fig. 8, APT 210 is made up of six subsystems; these in turn are composed of function Fig. 8 Illustration of functional structure of APT 210 SSS GSS TSS TCS CHS OMS AJ AJC AJR AJM Subscriber Switching Subsystem Group Switching Subsystem Trunk and Signalling Subsystem Traffic Routing and Control Subsystem Charging Subsystem Operation and Maintenance Subsystem A-Junctor AJ Hardware AJ Regional Software AJ Central Software blocks. Each function block is implemented in a number of function units, hardware or software. This functional hierarchy provides a well-arranged and easily handled structure. The interfaces between function blocks are strictly defined and chosen in such a manner that adaptation to different requirements, such as inter-exchange signalling and new facilities, affects only specific parts of the system. The subscriber switching subsystem, SSS, forms the interface towards the subscriber network. Functions for receiving subscriber signalling, for power feeding and ringing are located here together with the subscriber switch functions. 76 The trunk and TSS, forms the inter-exchange handles all the ling functions. signalling subsystem, interface towards the trunk network. TSS inter-exchange signal- The group switching subsystem, GSS, contains the group selector switching network together with functions for path selection. GSS is normally not affected by changes in the subscriber or trunk environment. GSS may be supplied in analogue or digital form. The traffic control subsystem, TCS, is implemented in software only. TCS contains the central routing and analysis functions for establishing and supervising each call and includes such functions as, for instance, automatic call forwarding. When a function of this type is required, the corresponding function block is included in TCS— the interface structure is already prepared. Fig. 9 APT 210 H a r d w a r e and software—function blocks for traffic handling. ( O M S not included) SSS SS AJ BJ KR SC Subscriber Subsystem Subscriber Line functions Subscriber Switching network functions A-Junctor (unctions B-Junctor (unctions Key-set Code Reception Subscriber Category analysis GSS GS Group Switching Subsystem Group Switching network functions TSS IT OT CR CS AS Trunk and Signalling Subsystem Incoming Trunk functions Outgoing Trunk (unctions Code Reception Code Sending Auxiliary Service (unctions TCS RA RE CL DA OF Traffic Routing and Control Subsystem Route Analysis Register functions Call Supervision Digit Analysis Optional Functions CHS AC CA MP MR TT Charging Subsystem Accounting Charging Analysis Metering Pulse generation Meter Reading Toll Ticketing The charging subsystem, CHS, contains functions for generating and storing appropriate metering pulses. Function blocks for toll ticketing may be included. Tariff determination data are accessible via the input-output system for resetting or additions. Subscriber meters are accessed for bulk or individual read-out. The operation and maintenance subsystem, OMS. The functions of this subsystem are prepared for centralisation of operation and maintenance. Traffic functions General The division of a subsystem into a number of function blocks is illustrated in fig. 9. The function blocks may be implemented in hardware and software or in software only. Each block is dedicated to a particular function such Fig. 10 APT 210 H a r d w a r e — t r a f f i c routing SSS SSN AJC BJC KRD LIC Subscriber Subsystem Subscriber Switching Network A-Junctor Circuit B-Junctor Circuit Keyset Code Reception Device Line Interlace Circuit GSS GSNI GSNO TSS Group Group Group Trunk ITC OTC CRD CSD ASD Incoming Trunk Circuit Outgoing Trunk Circuit Code Reception Device Code Sending Device Auxiliary Service Device Switching Subsystem Switching Network Incoming Switching Network Outgoing and Signalling Subsystem as incoming trunk, IT, digit analysis, DA, etc. The hardware portions of the function blocks of SSS, GSS and TSS are included in the trunking diagram, fig. 10, which illustrates the different traffic routes through an exchange. Function block To further illustrate the system structure, fig. 11 shows an example of a function block, the A-subscriber junctor, AJ. This block is made up of three parts: — the A-junctor hardware, AJC. This includes circuits for reception of dial pulses and on-hook/off-hook signals, for current feeding towards the calling subscriber, for sending of dial and busy tones, etc. — the A-junctor regional software, AJR, which scans and controls the AJC and performs signal recognition. — the A-junctor central software, AJM, which coordinates the traffic functions and the operation and maintenance functions of A J . As all block interwork is done by software signals, the hardware has obtain- Fig. 11 F u n c t i o n b l o c k AJ AJC AJR AJM Hardware of AJ Regional software of AJ Central software of AJ Fig. 12 An Extension Module, EM with 16 AJC 1 Printed circuit boards each with two AJC 2 EM control circuits ed a clearcut structure with very few and simple cable connections. Extension module, EM The APT 210 hardware is extended in modules, EM, each containing a number of identical devices or switches, fig. 12. The former is contained in a magazine with up to 16 devices, c o m mon circuits for test points, operation points and bus termination. Up to 64 EM are controlled by one d u p l i c a t e d regional processor, RP. The extension module, is the largest unit w h i c h can be affected by a single fault. Description of a traffic handling case To illustrate traffic handling, the simple example of an outgoing call will be described with reference to fig. 9. A call from a subscriber is signalled on the line circuit in LI. The subscriber c a t e gory is investigated and, if found to be a dial telephone, connection is made through SS to a free A J . (If the call comes from a pushbutton telephone, a further connection is made through GS to a free KR.) The digits sent by the subscriber are registered in AJ by scanning and stored in RE. 78 After reception and analysis of a sufficient number of digits, an outgoing trunk, OT, is selected and the path to it from AJ is reserved in GS. If the signalling system on the route requires a code sender, e.g. when MFC is used, a free CS is selected which, via GS, is connected to the OT. Digit transmission takes place from CS via GS and OT under the control of block RE: switching network between the subscriber switch subsystem and the group selector subsystem. As the current feeding and ringing functions are placed in the subscriber switch subsystem, no DC or ringing current is passed through the group selector, which may thus be implemented in analogue (reed switch) or in digital form. When an integrated register and line signalling system such as decadic pulsing is used, the signals are transmitted directly from OT under program control. This division has also enabled complete separation between the signalling media on the subscriber and trunk sides. Trunk signalling variations, for instance, will thus only affect the subsystem TSS, not GSS, SSS or TCS. After the signalling has been completed the CS link is released and AJ and OT are connected through the group switch; a check is also made that a speech path has been established through the switching network. During the call AJ and OT are supervised by the block CL. Simultaneously with the digit analysis, the charging subsystem, CHS, performs tariff analysis and metering. Switching network One of the characteristic features of the AXE system is the division of the Fig. 13 Basic analogue switchmodule in a three-stage reed switch number of inlets number of outlets size of the A-stage matrix size of the B-stage matrix size of the C-stage matrix number of inlets in switch module number of outlets or c-links in a switch module Switchblock groupings The basic analogue switchblock grouping is a three-stage reed switch arrangement, A, B and C in fig. 13. In these groupings there is always one and only one path between a given inlet and a given outlet—non-mesh network. The non-mesh network simplifies path selection; the system uses the network map principle. The basic switch module is a PC board with 64 crosspoint relays. This may be arranged in different ways to form 8X8, 8X4, 1 6 x 4 or other matrices. Fig. 14 b R e e d switch 1 2 3 Reed relay crosspoint Operating circuit Latching circuit F i g . 14 a Function D i a g r a m of reed switch consisting of a 2-wire r e e d relay crosspoint matrix " ^ ~ 1 2 3 Operation Holding Reed relay crosspoint Horizontal control leads Vertical control leads Reed switch — operation and holding The switch is a 2-wire reed relay crosspoint matrix and features a unique method of operation and latching using custom-designed integrated circuits. Fig. 14 a illustrates the principle and fig. 14 b shows a switch with 64 crosspoint relays with control logic. Table 1 Traffic capacity of SSN for different groupings To operate a crosspoint the control logic indicates the Y and X coordinates by opening the transistor Tr and the thyristor Th,. The crosspoint relay is operated in series with Th, from + 5 to —8 volts. The relay is held operated by a holding current from 0 volt in series with Th„ and Th,. To release the crosspoint relay the thyristor Th„ is shunted from the control logic. The thyristors are PNPN elements and the control logic consists of normal TTL decoders. A component board carries 64 crosspoint relays, one operating circuit and eight latching circuits. The outstanding feature of this 2-wire reed switch with electronic operation and holding is, apart from the saving of space, the immunity to disturbances. This is ensured by the use of thyristors; the operation and holding functions are not affected by, for instance, lightning-induced voltages (800 volts, 10 microseconds) or by the tripping of a 100 volts ringing signal. From a system point of view, for instance for test purposes, there are obvious advantages in having each crosspoint individually accessible for operation. Subscriber switching network, SSN Table 1 shows the matrices and corresponding traffic figures of the four different versions of the subscriber stage. The module is for 2048 subscribers. Transposition is obtained by a division of the network into two parallel planes. Analogue group selector, GSN GSN is made up of 3-stage blocks of 512 inlets and 768 or 1152 outlets. Two such blocks are connected together, forming a switching module which is controlled by a duplicated RP. A complete group selector consists of one or more such modules with the links (768 or 1152 per module) evenly distributed. This is a full availability, high performance group selector. The permissible traffic load per inlet has been chosen as high as indicated in table 2 in order to provide immunity against traffic overload and in order to eliminate the need for traffic equalization through rearrangement of inlets. Digital group selector, GSN-D APT 210 may be equipped with a digital version of the group selector GSN-D. GSN-D has the same interfaces as the analogue version; if digital lines are connected an additional digital hardware interface is introduced. The group selector consists of a digital switch in a TST structure (time-spacetime) plus PCM terminals as required, and digital line interfaces. For reliability the switch is fully duplicated so that two planes are formed. One of the planes can carry all the traffic; the loss is less than 10~6. GSN-D switches 8 PCM bits in parallel. Fig. 15 shows a diagram of an AXE 10 equipped with digital group selector. Operation and maintenance The major part of the operation and maintenance subsystem, OMS, is implemented in software. OMS supervises the operation of APT and takes appropriate action in case of disturbance. OMS includes some 30 function blocks, each of which is designed for a specific operation or maintenance task. The functions of OMS may be grouped according to the following headings. Supervision of traffic. One important feature of the AXE maintenance philosophy is direct supervision of live traffic. Faults or disturbances are detected and registered immediately they occur and corrective action is taken when necessary. An exchange thus does not have to carry the additional load of running routine test programs. Certain supervision functions are integrated into the traffic-carrying hardware; an example of this is the throughconnection check of the speech path for each connection. The supervision of devices such as junctors, trunks and signal devices is based on fault counters. When the num- Table 2 Max. number of inlets/outlets in the group selector GSN-A at different loads per inlet and different groupings. 2 °/oo internal loss ber of counts per time unit exceeds a preset quota indicating the probability that a fault has occurred, an alarm is given. Transmission measurement may be performed periodically or on command. Testing and fault location are initiated after receipt of alarm or subscriber complaints. OMS includes a number of test and fault analysis functions which may be started automatically or on command. Network supervision and traffic control. This feature is based on recording of handled traffic and, by means of blocking and rerouting strategies, assures that the trunk network is efficiently utilized and that overload situations are not allowed to spread through the network. Statistics. Modern operational practices are based on correct and up-todate knowledge of the traffic situation in all parts of the network. AXE 10 provides a wide range of recording functions, including traffic measurement and traffic supervision recording (number of lost calls). All results are presented as read-outs periodically or on command. Fig. 15 AXE 10 Switching system APT with digital group selector PCD GSN-D TSM SPM PCM Terminal Device Digital Group Switching Network Time Switch Module Space Switch Module Analogue Speech Transmission Digital Speech Transmission Administration. The traditional administrative functions of an exchange performed by strapping in different parts of the common control equipment are in AXE 10 replaced by working with commands via a typewriter, a display or a magnetic tape cartridge. This method is used for subscriber data (number group and categories) and for exchange data. The latter include the engineering information on routes and circuits, but also all the tables for analysis of route number, tariff, number length, etc. An additional category of special data are the individual lists for subscribers with abbreviated dialling facilities. These are handled by operators or, under certain conditions, by the subscribers themselves. Blocking of trunks or other devices is also performed by commands to function blocks in OMS. Centralized operation and maintenance. The structure of OMS with well defined standard interfaces towards the other subsystems facilitates centralization of operation and maintenance. OMS may be connected to an operation and maintenance centre OMC over a data channel, enabling all commands and read-outs to be handled centrally. 82 III. DATA PROCESSING SYSTEM APZ 210 The data processing system APZ 210 is designed with the aim of achieving simple handling, high reliability and a large call handling capacity. These goals have been attained by means of a structure that is well adapted to the basic characteristics of AXE. A number of small regional processors, RP, are connected to a powerful central processor, CP, via the regional processor bus RPB, see fig. 5. All interworking between the central processor and the regional processors is by means of messages (signals) transferred over the bus. There is no interworking between regional processors. F i g . 16 a Function units of c e n t r a l processor DS PS RS CPB RPB MAU CPU TCU RPH ALU BAM MIG PCU TRU DSH LIU PSH UPM RSH SBU Data Store Program Store Reference Store Central Processor Bus Regional Processor Bus Maintenance Unit Central Processing Unit Table and Counter Unit Regional Processor Handler Arithmetic and Logic Unit Maintenance Buffer Unit Micro Instruction Generator Priority Control Unit Trace Unit Data Store Handler Link and Instruction Address Unit Program Store Handler Updating and Match Unit Reference Store Handler Shift and Bit Handling Unit Fig. 16 b Central processor C P U RPB The regional processors handle the direct control of the connected equipment, such as switches, telephony devices and 10 (input/output) devices, whereas more complicated administrative functions are handled by the central processor. The central processor is duplicated, the two sides of the processor working in a synchronous parallel mode. Hardware faults can thus easily be detected by comparing the two sides. The regional processors are also duplicated, every telephony device being connected to two processors. If a fault occurs in one processor, the twin takes over the control of the devices. Central processor CP The central processor comprises the central processor unit CPU with associated stores, and also executive programs that administer job scheduling, program loading, storage allocation and program testing. The executive programs work in close cooperation with the hardware and provide the necessary service for the telephony programs. CPU is built up of a number of selfcontained function units that are connected to a common bus, see fig. 16 a and b. The work of the processor is controlled by a microprogram stored in a programmable read-only memory (PROM) in the microinstruction generator MIG. From MIG the other function units are ordered to carry out the desired operations. Microprogramming has made it possible to create powerful machine instructions adapted to the high-level programming language, PLEX. Auto- 83 nomous handlers in CPU interwork via buses with the RP's connected and with the main stores. These are of three types, viz. program store, PS data store, DS and reference store, RS. They are of similar design, using dynamic MOS as storage elements. RS constitutes a key for adaptation of APZ 210 to the modularity of AXE 10 and to the PLEX language. Combined with the register structure and CPU's microprograms efficient interworking between function blocks and high software security have thus been obtained. The central software part of a function block consists of programs with associated data. As mentioned before, the data can only be accessed by the programs in its own block. Interworking between different blocks is always carried out between programs in the respective blocks by utilizing special software signals. Only one block at a time can be active in the central processing unit CPU, i.e. the block having its block number temporarily stored in the block number register, fig. 17. This number gives access to the area in the reference store, RS, which belongs to the block. This area contains information both about the program start address and about the start address of the base address table for the block in question. The base address of a variable contains information on location, size, structure and other necessary characteristics. All address parameters in a [^Relative address parameters © Function block currently active © Instruction being executed © Data elements or variables Fig. 17 Disposition of central software stores © Instruction for accessing of a data element © Data individual being accessed 84 F i g . 18 J o b s c h e d u l i n g in CP PCU MFL TRL THL BAL MAU TRU JBA JBB JBC JBD EES Priority Control Unit M a l f u n c t i o n Level Trace Level Traffic Handling Level with sublevels 1—3 Base Level with sublevels 1 and 2 Maintenance Unit Trace Unit J o b Buffer A Job Buffer B Job Buffer C Job Buffer D End of Execution program are relative to the program start address or to the base addresses. This means, e.g., that an absolute address for a variable is obtained from a base address, which is specified by a relative address parameter in an instruction. This instruction is in turn indicated by the instruction address register. To reach individual data in a variable, the pointer register is used. tained. This greatly facilitates functional changes. A new feature is that the location of a program with associated data may be changed without changing its binary code. All software is thus relocatable. In case of a change it is thus only necessary to alter the block concerned, while the rest of the software is main- Job scheduling The job scheduling in CP is carried out mainly by microprograms and is based on an interrupt system with four program levels and a number of job buffers for queue administration, see fig. 18. As each block has its own base address table with individual data, it has access only to its own data. This simplifies the development and handling of the AXE 10 software to a great extent and provides a powerful protection against dispersion of faults. 85 The malfunction level MFL has the highest priority. An interrupt signal to MFL is generated when a fault detecting circuit has detected a hardware fault. Storage allocation Program blocks are read into the central processor stores by the executive programs, which then automatically carry out storage allocations. The frace level TRL is used in connection with program testing, see below. The executive programs also contain functions for changing the size of data areas in the data store. This facility is used when new subscribers or telephony devices are added to an exchange. Such an extension is prepared by the operator, who issues a command to the executive programs to increase the data store areas for the program blocks concerned. The increase is carried out automatically, and entirely without operational disturbances. When necessary, the executive programs also perform automatic storage packing to maintain good utilization. The normal work in CP is carried out at the traffic handling level THL and the base level BAL, with sublevels according to figure 18. BAL has the lowest priority. The job scheduling at THL and BAL is based on job buffers A—D for storing different signal messages. A signal message contains the block number and signal number for the receiving block and the data transferred with the signal. When a signal message is placed in a buffer, an interrupt signal is automatically generated. The priority rules are that interrupt signals to THL interrupt a job in progress at BAL but wait until a job in progress at THL is completed. An interrupt signal to BAL always waits until a job in progress has been completed. All signals from the regional processors and most signals between CP program blocks are placed in job buffers for intermediate storage. In addition, clock interrupt to program level THL takes place every 10th ms for transfer of signals to time-measuring program blocks in accordance with information in the job table. Program testing For testing purposes special microprogram-controlled tracer bits are activated. With these it is possible to trace signals between the program blocks and also changes in the program block data. When an activated tracer bit is encountered, an interrupt to the trace level takes place. The operator can order tracing with commands, so that interesting program sequences can be recorded and printed out. As the tracer bits are controlled by microprograms, the processor load is increased only insignificantly. Functional changes The executive programs allow substitution of program blocks and changeover from the old to the new programs without operational disturbances as long as the signalling interfaces between the blocks and their data structures are not changed. The duplication of the central processor may be utilized when making extensive functional changes in the software. One processor is then halted and loaded with the new software, after which a changeover is ordered so that the newly loaded processor takes over the operation. Operation is disturbed only insofar as calls in transition states are cut off. Connections that have already been established are not affected. Regional processor, RP A number of regional processors are connected to the central processor via the duplicated regional processor bus RPB. The regional processors perform simple routine tasks that require large capacity, such as scanning of test points and control of relays. The processor comprises a central processing unit, CPU, a data store, DS, a program store, PS, and executive programs for administering and supervising program handling. CPU consists 86 of a number of independent function units connected to a common bus, fig. 19 a and b. The work of the processor is controlled by a microprogram stored in a PROM memory. A number of extension modules, EM, are connected to the CPU via a bus EMB. An EM comprises a number of Fig. 19 a Regional processor RP Fig. 19 b Structure of regional processor RP EMB RPB CPU PS DS EMBU ALU PRU IBU MIG PSH DSH OBU RPBU Extension Module Bus Regional Processor Bus Central Processing Unit Program Store Data Store EM Bus Unit Arithmetic and Logic Unit Process Register Unit Input Buffer Unit Micro Instruction Unit Program Store Handler Data Store Handler Output Buffer Unit Regional Processor Bus Unit telephony devices or IO devices of the same type, or a part of the switching network. Two RP's control the same EM, though only one at a time. The RP program store is divided into a number of program pages. One program page is used for the RP executive programs. In addition one program 87 page is normally required for each type of device controlled by the processor. Each EM is allotted a continuous area in the data store. The start address of the data store area of an EM is stored in a base address table at the beginning of the data store. Job scheduling A program stored in a program page in an RP interworks with the corresponding program block in CP with signal messages over the bus, RPB. CP fetches the signal messages by continuously scanning all connected RP. All address translation in connection with the signalling (block numbers and device pointers in CP, EM numbers and internal device pointers in RP) is done automatically by the CP microprogram that administers the transfer of signals to and from RP. Programs are called in cyclically and this is administered by the executive program. The cycle period is usually 5 ms and constitutes the primary interval of the regional processor. During the primary interval the executive program scans all EM and calls in the corresponding programs in turn. Maintenance system The maintenance system in APZ 210 contributes to the high operational reliability that is necessary for telephone exchange applications. The maintenance programs work in close contact with the hardware and make use of built-in redundancy in the central and regional processors. Each central processor contains builtin fault detecting circuits of the following types: a) side-indicating supervisory circuits for, for example, parity check of stores and buses, as well as voltage and time supervision. In case of a fault these circuits directly indicate the faulty processor side. b) side-comparing supervisory circuits that detect faults by comparing data sent over the internal buses of the two central processors. c) program supervising circuits that detect, by time measurement, when the processor program handling does not operate normally. In addition, maintenance programs carry out routine checks of the central processor hardware. When a hardware fault is detected in CP, a fault signal is sent to the maintenance unit MAU. If a faulty processor side is indicated, MAU automatically halts this side and allows the other side to continue immediately with the traffic handling. If the fault is a comparison mismatch, MAU interrupts the work in progress and initiates a sideindicating program in both sides. The processor side that cannot execute this program correctly is halted by MAU. The interruption time necessary for identification of a faulty side is less than 20 ms. Thus hardware faults in the central processor are normally handled without traffic disturbance. A simple and reliable method of fault diagnosis is based on the fact that blocking and deblocking of the CP function units can be controlled by programs. When a processor side has been blocked after the detection of a hardware fault, the function units can be deblocked automatically, one by one, until a new fault signal is obtained, which directly and reliably pinpoints the faulty unit. A number of program operation checks have been incorporated in the microprograms, which' has been facilitated by the structure adopted for APZ 210. This assures early detection of faults and minimizes fault dispersion, thus securing a high software quality. The regional processors are duplicated so that each processor in a pair normally controls half of the common EM's. Each RP has its own supervisory circuits for parity checks and time supervision. In addition, both central and regional routine test programs are provided for the supervision of RP. 88 When a fault is discovered in an RP, the work in progress is automatically interrupted and a fault signal is sent to the maintenance programs in CP. Depending on the extent of the fault CP then transfers the control of the EM's concerned to the regional processor that is not faulty. way a simple command is written is given below: For certain serious faults a system restart is carried out, which means that traffic handling is resumed from a well defined starting point. Print-outs can be guided to various devices in a simple way since all printouts are classified with a number of function codes. By means of a command the operator can guide print-outs belonging to a certain class to one or more given IO devices. If a number of system restarts take place within a short time, all software is automatically reloaded from magnetic tape. However, a system restart usually affects only calls that are in process of being set up, while established calls are not affected. 10 system (IOS) Fig. 20 4 K9 memory board The 10 system comprises functions for transfer of information between APZ 210 and the operator or other computer systems. During the design work the most important goals were to achieve reliable man-machine communication and good flexibility for future technical developments. The 10 system is divided into blocks in accordance with the AXE 10 function block structure. The IO system contains a number of administrative program blocks and also a function block for each type of IO device. The device blocks contain central and regional software and the corresponding hardware. The regional software carries out the direct control of the devices while the central software handles administration and processing of information. A special command language has been developed for AXE. When designing the language, regard was paid to the standardisation work in progress within CCITT. As a result of the logic structure of the language, the analysis of commands is to a great extent done in the IO programs. This prevents the user programs being affected by any syntactic changes. An example of the LAPBL: BLOCK = ABC, IODEV = C T 1 ; The command orders the read-in of program block ABC from magnetic tape cartridge reader no. 1, Conclusion The two most important features of the AXE 10 system are the functional modularity with integrated hardware and software and the provision of alternative analogue and digital switchblocks in the group selector. The functional modularity is the basis for obtaining a system with substantially improved handling properties. Functional modularity is also the means for software security, the characteristic of the new generation SPC. AXE 10 offers a very wide range of subscriber services and, equally important, the administrative facilities that open the way to centralization of operation and maintenance. AXE 10 is thus ideally suited both for today's modern networks and for the networks of tomorrow, the networks with an increasing proportion of integrated digital switching and transmission and with highly sophisticated, centralized operation and maintenance. 89 Technical data for APZ 210 COMPONENTS AXE 10-Software Structure and Features Goran Hemdal This paper describes the structure of the AXE system with emphasis on the characteristic functional modularity. This unique structure provides the basis for the two outstanding features of the system—software security and ease of handling. U D C 621.395.722: 681.3.065 681.3 06 L M E 83022 Fig. 1 E x t r a c t f r o m the p r o d u c t s t r u c t u r e of t h e AXE s y s t e m SSS AJ AJC AJR AJM RP CP Subscriber subsystem A-junctor functions AJ c i r c u i t AJ regional software AJ central software Regional processor Central processor Experience gained from SPC systems of the generation at present in service has shown that the problems encountered in handling the software of the systems have been greatly underestimated. When developing the AXE system software one of the basic reguirements was therefore the elimination of the bulk of traditional software problems by creating a system in which software handling is simple and which at the same time provides high operational security. One of the most important features of the AXE software is its modularity, which directly corresponds with the functional modularity of the system. This modularity allows every software module to be independently programmed while at the same time ensuring that the modules will work together. As the modules are small and a high/ level language is used for the programming, a very high program quality is obtained. As, furthermore, the functional modularity is maintained in the implementation of the software, every software module may be separately compiled and loaded. Storage allocation is performed as part of the loading activity on the target processor. This will facilitate the software handling, especially under normal service conditions, for instance for functional changes, when any disturbance of the normal service is avoided. Software structure Functional module and product In order to provide ease of handling in combination with high software security the AXE system software is integrated into the functional modules of the system. Each such module (system, subsystem, function block or function unit) is defined as a product. Together the products form the product structure of the AXE system, fig. 1. Viewed from the aspect of the remainder of the system, each such individual product is considered a "black box" with unambigously defined interfaces with interworking products. The interface contains all the different signals which the product may send and receive during operation. A signal signifies the transfer of information, at this stage without regard to the method of transfer. For every product the interface of the product is strictly defined, enabling the product to be used as a building block for higher-level products. Every product of the system, subsystem and function block level is assembled from the products on the next 91 GORAN HEMDAL ELLEMTEL lower level. A specification of these products therefore contains only references to the appropriate specifications of the products on the next level. The interconnection of the interface of every subordinate product to the interface of every other subordinate product, as well as to the interface of the product itself, is however specified in detail. In order to enable automatic processing of engineering activities, such as determination of the products involved for a particular AXE exchange, a product specification also defines the necessary exchange assembly parameters. For every parameter the method of passing the value of the parameter between the products on different levels is specified. By means of these exchange assembly parameters, Fig. 2 Example of two functional alternatives for one product functional alternatives may be specified. Fig. 2 shows two examples of specification of function block X. The total product structure for the AXE system defined by the product specifications enables an automatic translation of the functional requirements of any particular exchange to the appropriate product configuration satisfying these requirements. The strictly defined interfaces and their equally strictly defined interconnections ensure that every product configuration obtained is also a working configuration. A product specification on the function unit level differs from the specifications on the higher levels with respect to the internal structure, which is specified in relevant hardware or software terms. 92 Hardware function unit This unit consists of a number of identical devices, fig. 3, for instance trunk circuits, code receivers, etc. Interworking with other connected hardware function units is performed by hardware signals, that is to say the information is transferred by, for instance, pulses or combinations of frequencies on a physical transmission path. Software function unit This consists essentially of a number of identical "devices" which may be divided into a program part and a data part. The program part consists of a sequence of program instructions— the program logic of the device, while the data part contains information about the state of the device, linkages to other devices, etc. These data are stored in data elements of lengths suited to the information. In the AXE system each such data element is defined as a variable. Fig. 3 Principal structure of a hardware function unit with n devices Fig. 4 Principal structure of a software function unit with records corresponding to n devices Both the program part and the data part of a software function unit are stored in the memories of the processor which executes the logical functions. As it is unnecessary to duplicate identical information, the program parts of all de- vices form one common program. In the same way, sections of the data parts are combined in common data, while data that are individual for each device form individual data, fig. 4. The variables, which form the common data of the function unit, are directly accessible from the program. For individual data, on the other hand, the variables are organized in a record for each device. To reach an individual variable the device in question must be indicated first. In AXE the necessary indication mechanism is defined as a pointer. No detail of the program and data structure of a software function unit is accessible from outside, all interwork with other function units is performed by signals. As a consequence no globally accessible data exist in the system. Every software function unit can therefore be programmed independently, while the product structure ensures that the function units will interwork correctly. Program errors are effectively prevented from mutilating data belonging to other function units. As a consequence both programming and debugging are simple. 93 Software signals As the information carried by the signals between a software function unit and interworking function units is transferred by software means, these signals are referred to as software signals. Transmission of a software signal is specified as an instruction in the program. This instruction associates the variables and constants of the sending function unit with the information Fig. 5 format of the signal. In this way the values of the variables and constants are transferred as signal information although the variables and constants themselves are accessible only within the sending function unit. Reception of a software signal is also specified as an instruction. The function of this instruction is twofold. The instruction as such indicates the location in the program where program execution is to start when the signal identified in the instruction is received. Secondly the instruction asociates the information format of the signal with the variables of the receiving function unit, thereby making the signal information accessible. Principles of function unit interworking Software signals are always interchanged with the help of the processor, fig. 5. This processor may transfer the signal to another software function unit executed by the same processor, to another software function unit executed by another processor or to a hardware function unit. The necessary transformation from software signals to hardware signals is performed by the processor. In the same way hardware signals are transformed to software signals when transferring signals from a hardware to a software function unit. Hardware signals Software signals 94 Software implementation Implementation principles From the customer's point of view the characteristics of a system during normal service are far more important than the characteristics during specification and design. In order to obtain the manageability and software security required of the AXE system, the functional modularity must be maintained in the implementation of the system. One possible way of achieving this would be to provide each software function unit with its own processor. For technical and economical reasons this is not feasible today. It is however possible to design a processor so that, although a number of function units exist in the same processor, this processor provides exactly the same characteristics for each function unit as if it had its own processor. Fig. 6 Function block structure An analysis of the software functions of an SPC system leads to a division into highly frequent simple functions and complex functions of low frequency. For technical and economical reasons, therefore, it was advantageous to structure the data processing system APZ 210 of the AXE system as a hierarchy with two types of processors—central and regional processors. The central processor is designed to execute the low-frequent complex functions, while the regional processors are geared to the performance of repetitive simple functions. There are thus three types of function units in the AXE system—hardware function units, regional software function units and central software function units. The regional software function units are used, for instance, for scanning test points and operating relays in hardware function units and can only interwork with hardware function units or central software function units within the same function block. Interworking between different function blocks is as a rule performed by software signalling between central software function units. A function block can thus consist of any combination of central software function units, regional software function units and hardware units, fig. 6. As it must be possible to choose the most suitable combination of implementation techniques for any function, ii._ function blocks have been selected as the normal handling products. Both the central and the regional processors of the APZ 210 system are therefore geared to the function blocks. Implementation of central software Only one function block at a time may be active in the central processor. For this reason each function block is assigned a unique block number. The central software of a function block is active only when its corresponding block number is found in the central processor block number register, fig. 7. With the aid of the block number a fixed reference table in the reference store of the central processor is accessed. The following information relating to the function block iscontained in this table: —• the state, e.g. normal operation — the program start address, giving the absolute address to the area where the programs of the central software unit belonging to the block are stored — the base start address, providing start address of the area where the base addresses for the variables of the central software unit belonging to the block are stored. The program of the function block begins with a signal distribution table. Each signal to a block is identified by a signal number which is unique within the block. The number of a received signal is translated to the correct entry position with the aid of the signal distribution table. The base address contains, among other things, the absolute address of the variable in the data store, the length of the variable, information as to whether the variable belongs to common or individual data a n r i in the- l a t t e r racp. 95 the number of devices. When an arbitrary variable is accessed, only the number of the variable within the function block base address table is given in the instructions, whereafter the microprograms of the central processor perform the necessary address calculations. for every function block may be compiled separately, while at the same time the machine code obtained from the compilation is fully relocatable. All memory allocation is therefore performed by the central processor itself in connection with the loading of the central software. At the same time the microprogram checks that the resulting absolute address in the data store really is situated within the area assigned to the current variable. Each variable thus has its individual memory protection. As accessing of variables belonging to other function blocks is prohibited by the central processor hardware, and as each variable has its own individual memory protection, powerful protection is obtained against the dispersion of software faults. At the same time the probability of locating program faults already during system testing has been substantially increased. By designing the machine instructions in such a way that they always refer to quantities in relation to current start addresses, the central software Fig. 7 Implementation of central software BS PSA BSA Block State Program Start Adress Base Start Adress 96 Implementation of regional software A regional processor stores and executes the regional software in the same way as the central processor. In this case the procedure is much simpler as there is no interworking between different function blocks in the regional processor. The regional software of a function block is stored with the program in a program page in the program store of the regional processor, and the data is stored in a continuous area in the data store, fig. 8. There are in this case no base addresses as in the central processor, but as the variables within a record form a continuous area where each variable is defined by its position within the record, the same advantages are obtained as in the central processor. Software representation Source documents The software of the AXE system is represented in the form of source documents. These source documents specify the different types of software information used in the AXE system and they consist of signal documents, function unit documents, product structure documents and project documents. Fig. 8 Implementation of regional software The signal documents specify each software signal as to its properties, the format of the transferred signal information, etc. The function unit documents specify the internal structure of each function unit, i.e., among other particulars its data structure and program logic. The product structure documents specify the internal structure of products above the level of the function unit. The project documents specify the selection of a c o m b i n a t i o n of working products for a specific exchange taken from the total collection of the AXE system as specified by the product structure and function unit documents. Current source documents are handled with the aid of the programming system APS 210. The object is of course to obtain, in target code form, the software of a specific exchange. Language family APS 210 defines necessary source document languages in the form of a language family, w h i c h includes the special source languages SPLEX, PLEX, ASA 210C, ASA 210R and CLAPS, fig. 9. 97 SPLEX includes facilities for declaration of signals and the internal structure of the products, while PLEX c o n tains facilities for specification of program logic. PLEX is a high-level, machine-independent programming language for SPC systems and, apart from normal highlevel language facilities, includes facilities for signal sending and reception and pointer handling. A S A 2 1 0 C and ASA210R are the respective assembly languages for the central and regional processors of APZ 210. CLAPS includes the necessary control and editing directives for handling APS 210 in the design, engineering, production, etc. of the AXE system software. Software production The application programs are initially written in the high-level language PLEX and their formal correctness is analysed by the APS 210 programming system. A check is then made that the signal reception and transmission statements match the corresponding signal formats specified in the signal documents written in the SPLEX language. This is an absolute requirement before any software function unit may serve as a building block. The programs are also tested on the source code level (PLEX), thereby eliminating almost all program errors at an early stage. The p r o g r a m m i n g system APS 210 provides the necessary printouts of the source code as well as of analysis and test results, fig. 10. When producing the object software for APZ 210 in a specific exchange, the exchange assembly parameter values contained in the project document for this exchange are used for specifying the actual functional requirements. With the aid of the product structure documents these values are translated into a list of all products to be included in the exchange. At the same time the block and signal numbers are assigned. APS then compiles the software for each function block separately into object code, which is normally stored on cassette tape. The same cassette tape may contain one single function block or any number of such blocks. A printout of the object code is also obtained, in which every instruction is listed both in binary machine (object) code and in ASA210C assembly code, fig. 10. AXE Fig. 9 Software languages and their usage SPLEX PLEX ASA210C ASA210R CLAPS Specification Language for Exchanges Programming Language for Exchanges Assembly language of the APZ 210 Central processor Assembly language of the APZ 210 Regional processor Control Language for APS 98 Exchange-dependent data are not specified in the source code and will consequently not be included in the object code. Instead, the exchangedependent data are loaded separately with normal commands. coding in PLEX. As APZ 210 is adapted to the functional structure of SPC systems, efficient machine code is generated from programs written in PLEX. Only purely machine-dependent parts of the operative programs in APZ 210 are written in assembly code. Software characteristics program security. The functional structure of APZ 210 combined with the addressing principle provide an efficient limitation to inadvertent programming faults. Such faults are consequently few and easy to find during program testing. The functional orientation in the structure of the AXE system software, combined with the functional structure at target code level of the data processing system APZ 210, provides the software with a number of important features such as — structured system engineering. For system design a functional structuring is performed, in subsystem, function block and function unit. Programming is done only at the function unit level. Fig. 10 Software production activities high program quality. The principles of program design as listed in the points above, plus the use of sophisticated aids, such as the programming system APS 210, for design, testing and production of software, guarantee a high quality of the AXE system software. 99 — function blocks compiled separately. Because of the functional structureeach function block iscompiled separately. This means that, for example when correcting possible program faults, a new version of the function block is compiled and exchanged for the faulty one. Thus "patches" are avoided. — functional blocks loaded separately. As all target code is relocatable in the memories of the data processing system, all memory allocation is done in connection with loading the software on the target machine. This means that, for example, most functional changes are performed during normal service, without traffic disturbance. — exchange data loaded by commands. All exchange data are loaded by commands. This means that all types of extensions are effected directly on site without the need of an administrative computer. Fig. 11 Loading of a cartridge unit — functional signal tracing. All signals between the different function blocks of the AXE system may be traced. This enables the maintenance personnel to locate functional faults to the faulty unit without having to use the detailed software documentation. Conclusion The functional modularity of the AXE system, which is maintained in the implementation of the system, gives the AXE system its outstanding flexibility, manageability and software security for all phases of design, engineering, production and installation, as well as during normal service. With the AXE 10 system it has therefore been possible to attain all the original objectives of the SPC technique. New Packaging Structure tor Electronic Switching Equipment Rune Alexandersson and Hans 0. Rorstrom The design of a new mechanical packaging structure for electronic switching equipment was initiated at the same time as the AXE system project was started. This meant that the functional concept of the AXE system influenced the packaging structure. As a matter of fact the mechanical structure should be seen as an integral part of the AXE system. It contributes to a very large extent to the outstanding AXE system features which provide ease of handling—in production, design, testing, installation and extension. This paper gives a brief presentation of the main characteristics of the mechanical structure. UDC 621 395 722: 681.3.065 621.3572 LME 738 83022 The basic requirement for the design of the new packaging structure was to provide a suitable structure for predominantly electronic equipment with components of the IC type, but also with miniature electromechanical components and various special devices, such as power packs, fig. 1. At the same time it should be recognized that modern system design, documentation and production impose a particular set of requirements because of the use of computerized aids in all the different handling processes. These technical requirements are naturally coupled to certain economical guidelines. In the present case the main emphasis has been on the minimization of total cost, including installation and operation, rather than on Fig. 1 AXE exchange. The mechanics are painted in two shades ot grey—a basic light grey and a darker matt shade for end covers and cable section doors; power converters and markings are blue trying to develop a packaging structure that might seem economical as a subsystem but would not permit the exploitation of the economical potential of the total system. Total cost minimization has been reached by designing the packaging structure in such a way that the functional structure of the AXE system is maintained in the mechanical design. In practice this means that the functional module of the AXE system—which is also the module used for design and documentation—is identical to the mechanical module, the magazine. This unique property of the integrated system has meant that the traditional rack, with its wiring, testing and documentation, has been eliminated—the main unit is the magazine corresponding to the hardware of the function block. Requirements The mechanization and standardization of the handling activities in design and production also mean corresponding handling improvements for the telephone administration. For instance operation and maintenance are simplified, as also are planning and execution of all forms of extensions or functional changes. However, the extensive use of computer-based handling aids 101 RUNE ALEXANDERSSON HANS O. RORSTROM Telephone Exchange Division, Telefonaktiebolaget LM Ericsson in large-scale production must not exclude the possibility of producing the AXE system in local factories or running it in networks largely using conventional methods. Examples of typical activities which to a great extent are computerized are printed board design and production, production of wiring tables for magazine backplanes, production of testing programs and cabling tables. Table 1 UNITS OF MEASURE Basic unit Building module 1 M = 2.54 mm (M = Module) 1 BM = 16 M = 40.64 mm For production tooling the mechanical design has many advantages, as the structure is very simple. Production methods are thus easily adjusted both to the almost wholly manual handling of short series and to the fully automated production techniques for long series. Local production of the packaging structure can therefore easily be arranged. As the structure is modular, so that a large number of units are composed of a basic assortment of a few mechanical details, the spare parts handling as well is simplified. Functional structure The functional hierarchy of the AXE system is seen in fig. 2, showing the four levels: system, subsystem, function block and function unit. The hardware structure is divided up in the same way. The function block with its documentation corresponds to the magazine, which forms the handling module, with its documentation. The function blocks vary in size and therefore it is necessary to provide a series of different magazine sizes. In traditional rack design this would not be feasible; instead a shelf structure has been developed in which magazines of varying sizes are mounted— there is obviously a very large freedom of choice of combinations. Fig. 2 Functional hierarchy of the AXE system (Blue) hardware (Yellow) regional software (Red) central software The advantages of these new concepts in telephone exchange construction are apparent in all phases of handling. In production these units, magazines, are equipped with their printed circuit boards and tested as functional units. They are then packed and shipped, with the PC boards in position, and arrive on the installation site as fully tested functional units—one result of this practice is considerably reduced installation handling, especially testing, and appreciably reduced installation lead times. The use of different combinations of mechanical piece-partsout of the basic assortment is governed by design rules based on modular concepts. The packaging structure is adapted for both electronic and electromechanical components. The unit of measure for the basic module is 1 M = 2.54 millimetres. The building module 1 BM = 16 M = 40.64 mm has been introduced to facilitate handling of the larger modular measures, table 1. The designations for the different parts of the structure will be found in table 2. The designation BYB 101 is also used collectively to denote the complete packaging structure. The magazine is defined as a unit which contains a number of printed board assemblies and also interconnects the boards electrically. The row structure mechanics constitutes all the mechanical piece-parts used to form a row with its shelves, cable chutes etc, which accommodate the magazines, cabling and so on. Characteristics of the packaging structure The features of the BYB 101 structure, as described and used in a functionally modular' way, may be summarized in the following points: • • • • • • • • • • common mechanical, electrical and functional interfaces documentation adapted to handling racks replaced by magazines flexible grouping of magazines cabling separated from and independent of the row structure standardized "connectorized" cables all cables connected from the front decentralized power distribution good heat dissipation properties attractive and modern external design. 102 These points are explained and discussed briefly in the following paragraphs. Common mechanical, electrical and functional interfaces This means that a certain electrical function corresponds to a mechanical unit. This correspondence has advantages in all stagesof handling of the Fig. 3 PC Board magazines. The magazines are made up of two main components—the board trame(s) and the back plane (wiring unit) Fig. 4 Printed board assemblies product. The various handling activities are design, documentation, production, testing, shipping, installation, and operation and maintenance. This common interface is obtained by the modular flexibility of the mechanics, providing a range of magazine sizes to suit different functions. There is a series of magazines for from 5 up to 62 printed boards, figs. 3 and 4. 103 I Fig. 5 Documentation adapted to handling Documentation adapted to handling This is a natural consequence of the coinciding of interfaces. There are in principle no documents that extend outside the handling module, the function block. A documentation structure, defined per type of product, regulates the types and numbers of documents necessary for each unit. This structure increases the possibilities of data processing of the documentation. The use of new documentation media is foreseen, for instance, in the form of centralized data bases, fig. 5. Racks replaced by magazines The traditional rack in the form of a vertical structure with its own documentation, wiring, testing, etc. has been eliminated in the BYB 101 equipment practice. Today the board magazine is the major mechanical, wiring and functional unit. Fig. 6 Row structure principle The row structure is a new concept. The total length is decided by the exchange size and application and by the available space. The row is subdivided into smaller units, called sections, of two types—shelf section and cable section. The shelf section consists of horizontal shelves on which the magazines are placed. The shelves also carry the horizontal cables to the magazines. The cable section is a vertical unit in which the cables are run from the cable chutes along the top of the row to the respective shelves of the shelf section. The row may be single- or double-sided, fig. 6. The different combinations of shelf and cable sections are practically unlimited. Thus it is always easy to adapt the mechanical structure to any desired size or application. 104 The row mechanics is a standard product which is delivered from stock directly to the installation site, fig. 7. Fig. 7 Row structure mechanics, main components. The verticals form the framework of both shelf and cable sections. The front and back covers are used for protection and air guiding and also form an attractive exterior Fig. 8 Flexible arrangement of magazines— horizontally and vertically Fig. 9 Modular building elements of a magazine. (From left) wiring unit width 15 BM, board frame 3 BM and board frame 9 BM Flexible grouping of magazines Magazines may be placed beside each other, horizontally, or above one an- other, vertically, like books on a shelf. It is thus easy to find the particular grouping which requires minimum cable lengths and floor space. Alterations or extensions are also easy to engineer and execute—magazines can be replaced or moved without change of the row structure mechanics, fig. 8. 105 11 L -i. L Fig. 10 Cabling separated from row structure Cabling separated from row structure As mentioned, the row structure includes no cabling; all connections between magazines are made with plugin cables, one of the prerequisites for flexible magazinegrouping. Installation is simplified as separate documentation is used for the mechanics and for cabling. The row structure mechanics is system-independent, fig. 10. Standardized "connectorized" cables The number of different cable types has been restricted in order to simplify cable handling. The cables are always manufactured on the one-to-one prinprinciple, i.e. pin 01 at one end is connected to pin 01 at the other end, pin 02 to 02, etc., fig. 11. Fig. 11 Standard plug-in cables with pin-to-pin interconnection Fig. 12 Below Bus cable. For 24 and 16 pair buses Fig. 13 Signal cables, with connectors. (From top) 32, 16, 12, 8 and 4 pair cables The cabling principle results in a minimum of contacts in each wiring con- nection and thus a minimization of possible fault points. Connections between two circuits (printed boards) are always direct. The standardization of the cables has enabled simplifications in documentation, production and testing. To a great extent prefabricated cables can be used, factory-produced, delivered complete and tested on site. There are four main types of cable: n signal cable • signal cable with test facility • bus cable • power distribution cable Figs. 12 and 13 show examples of different cables. 106 All cables connected from front All cables are connected from the front of the row to board front connectors or to a connection field in the wiring unit. In this way existing cables may be changed or new cables added and additional equipment is easy to install. The front connection also aids operation and maintenance work, as it is easy to connect test equipment and instruments, fig. 14. The earthing system has, where possible, been incorporated in the structure to ensure perfect exchange earthing. The distribution system has socalled dynamic earthing, using an earth lead parallel to the power lead. This provides an equalizing effect on the occasional transients2. Decentralized power distribution The voltages necessary to drive the electronics are produced locally in the respective magazines. This structure forms part of the reliability strategy of the systems. Power is provided by DC/DC converters; normally these are not duplicated, fig. 15. packing density, however, heat dissipation often sets a limit to packing. Good heat dissipation properties Component minimization enables high Fig. 14 Front connection. Cables connected to PC board fronts and to wiring unit Fig. 15 Decentralized power distribution. 50 W converter in magazine The largest, 200 W, converter forms a unit, which is handled as a magazine, fig. 16. In normal conditions the AXE system requires no forced air cooling. The heated air which is funnelled up through the structure should be removed from the switchroom by extraction ventilators, fig. 17. In severe climatic conditions air coolinn and fans mav be installed. 107 Technical data UNITS OF MEASURE Basic unit: 1 M = 2.54 mm (M = Module) Building module: 1 BM = 16 M = 40.64 mm ROW STRUCTURE MECHANICS BYB 101 Height: 2250 mm (6 shelves) or 2900 mm (8 shelves) Depth: Single row: 360 mm Double row: 620 mm Shelf section width: 18 BM (731.52 mm) 24 BM (975.36 mm) Cable section width: 3 BM (121.92 mm) 6 BM (243.84 mm) Fig. 16 DC/DC converters. There are different sizes for 200 W, 50 W and 20 W. Floating output is used so that negative as well as positive voltages can be produced MAGAZINE BFD 111—129 Height: 6 BM (243.84 mm) Width: N X 3 B M , where N = 1 , 2 , 3 . . . 8 Depth: 220 mm Board spacing: 6 M (15.24 mm) or 8 M (20.32 mm) Number of PC boards: From 6 up to 62 PCB assemblies of type ROF13 * Only for centralprocessor unit APZ 210 ** Only as daughter-board for joint printed board CONNECTOR According to IEC Draft 48 B (Germany) 53 Standard 32 pins, pin spacing 2 X 2 M Special 48 pins, pin spacing 1 X2 M References 1. Fig. 17 Parallel air cooling through free convection Sorme, K. and Jonsson, I.: AXE — a functionally modular SPC system. ISS 74 — 411/1—7. 2. Orevik, A.: Power supplies for electronic telephone exchanges. Ericsson Review 57 (1974): 4, pp 120—127. The Ericsson Group SM With associated companies and representatives EUROPE SWEDEN Stockholm 1. Telefonaktiebolaget LM Ericsson 2. L M Ericsson Telemateriel AB 1. AB Rifa 1. Sieverts Kabelverk AB 1. Svenska Radio AB 5. ELLEMTEL Utvecklings AB 1. AB Transvertex 4. Svenska Elqrossist AB SELGA 1. Kabmatik AB 4. Holm & Ericsons Elektriska AB 4. Mellansvenska Elektriska AB 4. SELGA Mellansverige AB Alingsas 3. Kabeldon AB Gothenburg 4. SELGA Vastsverige AB Kungsbacka 3. Bota Kabel AB Malmo 3. Bjurhagens Fabrikers AB 4. SELGA Sydsverige AB Norrkoping 3. AB Norrkopings Kabelfabrik 4. SELGA Dstsverige AB Nykoping 1. Thorsman & Co AB Sundsvall 4. SELGA Norrland AB Uddevalla 4. Wamebolaget AB EUROPE (excluding Sweden) DENMARK Copenhagen 2. L M Ericsson A/S 1. Dansk Signal Industri A/S 3. GNT AUTOMATIC A/S FINLAND Jorvas 1. Oy L M Ericsson Ab FRANCE Paris 1 . Societe Francaise des Telephones Ericsson 2. Thorsmans S.A.R.L Boulogne sur Mer 1. RIFA S.A. Lannione 6. Societe Lannionaise d ' E l e c t r o n i q u e SLE-CITEREL Marseille 2. Etablissements Ferrer-Auran S.A. IRELAND Drogheda 1. L M Ericsson Ltd. 1. Thorsman Ireland Ltd. PORTUGAL Lisbon 2. Sociedade Ericsson de Portugal HAITI Port-au-Prince 7. Telefonaktiebolaget LM Ericsson SPAIN Madrid 1. Industrias de Telecomunicaci6n S.A. (Intelsa) 1. L M Ericsson S.A. MEXICO Mexico D.F. 1. Teleindustria Ericsson, S.A. 1. Latinoamericana de Cables S.A. de C V. 2. Telefonos Ericsson S A 2. Telemontaje, S.A. de C V Lda SWITZERLAND Zurich 2. Ericsson AG UNITED KINGDOM Horsham 4. Thorn-Ericsson Telecommunications (Sales) Ltd. 2. Swedish Ericsson Rentals Ltd. 5. Swedish Ericsson Company Ltd. London 3. Thorn-Ericsson Telecommunications (Mfg) Ltd. 6. Thorn-Ericsson T e l e c o m m u nications Ltd. 4. United Marine Leasing Ltd. 4. United Marine Electronics (UK) Ltd. WEST GERMANY Hamburg 4. UME Marine Nachrichtentechnik GmbH Hanover 2. Ericsson Centrum GmbH Ludenscheld 2. Thorsman & Co GmbH PERU Lima 2. Cla Ericsson S.A. EL SALVADOR San Salvador 7. Telefonaktiebolaget LM Ericsson URUGUAY Montevideo 2. Cia Ericsson S.A. VENEZUELA Caracas 1. Cia An6nima Ericsson Representatives In: Bolivia, Costa Rica, Dominican Republic, Guadeloupe, Guatemala, Guyana, Haiti, Honduras, Netherlands Antilles, Nicaragua, Panama, Paraguay, El Salvador, Surinam, T r i n i d a d , Tobago. Representatives In: Austria, Belgium, Greece, Iceland, Luxembourg, Yugoslavia AFRICA LATIN AMERICA ARGENTINA Buenos Aires 1. Cla Ericsson S.A.C.I. 1. Industrias Electricas de Quilmes S.A. 5. Cla Argentina de Telefonos S.A. 5. Cla Entrerrianade Telefonos S.A. BRAZIL Sao Paulo 1. Ericsson do Brasil Comercio e Industria S.A. 4. Sielte S.A. Instalagoes Eletricas e Telefonicas 4. TELEPLAN, Projetos e Planejamentos de Telecomunicagoes S.A. Rio de Janeiro 3. Fios e Cabos Plasticos do Brasil S.A. Sao Jose dos Campos 1. Telecomponentes Comercio e Industria S.A. ITALY Rome 1. FATME Soc. per Az. 5. SETEMER Soc. per Az. 2. SIELTE Soc. per Az. CHILE Santiago 2. Cla Ericsson de Chile S.A. NETHERLANDS Rijen 1. Ericsson Telefoonmaatschappij B.V. COLOMBIA Bogota 1. Ericsson de Colombia S.A. Call 1. Fabricas Colombianas de Mater i a l s Electricos Facomec S.A. NORWAY Oslo 3. A/S Elektrisk Bureau 2. SRA Radio A/S 4. A/S Telesystemer 4. A/S United Marine Electronics PANAMA Panama City 2. Telequipos S.A. 7. Telefonaktiebolaget LM Ericsson COSTA RICA San Jose 7. Telefonaktiebolaget LM Ericsson Drammen 3. A/S Norsk Kabelfabrik ECUADOR Quito 2. Telefonos Ericsson C.A. POLAND Warszaw 7. Telefonaktiebolaget LM Ericsson GUATEMALA Guatemala City 7. Telefonaktiebloaget LM Ericsson ALGERIA Algiers 7. Telefonaktiebolaget LM Ericsson EGYPT Cairo 7. Telefonaktiebolaget LM Ericsson MOROCCO Casablanca 2. Societe Marocaine des Telephones et Telecommunications Associee au Group Ericsson •SOTELEC" TUNISIA Tunis 7. Telefonaktiebolaget LM Ericsson ZAMBIA Lusaka 2. Ericsson (Zambia Limited) 2. Telefonaktiebolaget LM Ericsson Installation Branch Representatives In: Angola, United Arab Emirates, Cameroon, Central African Republic. Chad, People's Republic of the Congo, Dahomey, Ethiopia, Gabon, Ivory Coast, Kenya, Liberia, Libya, Malagasy, M a l a w i , Mali, Malta, Mauretania, Mozambique, Namibia, Niger, Nigeria, Republic of South Africa, Reunion, Senegal, Sudan, Tanzania, Tunisia, Uganda, Upper Volta, Zaire. ASIA INDIA Calcutta 2. Ericsson India Limited INDONESIA Jakarta 2. Ericsson Telephone Sales Corporation AB IRAQ Baghdad 7. Telefonaktiebolaget L M Ericsson IRAN Teheran 7. Ericsson Telephone Sales Corporation AB KUWAIT Kuwait 7. Telefonaktiebolaget L M Ericsson LEBANON Beyrouth 2. Societe Libanaise des Telephones Ericsson MALAYSIA Shah Alam 1. T e l e c o m m u n i c a t i o n Manufacturers (Malaysia) SDN BHD OMAN Muscat 7. Telefonaktiebolaget LM Ericsson THAILAND Bangkok 2. Ericsson Telephone Corporation Far East AB TURKEY Ankara 2. Ericsson Turk Ticaret Ltd. Sirketi Representatives In: Bahrein. Bangladesh, Burma. Cambodia. Cyprus. Hong Kong. Iran. Iraq. Jordan, Kuwait, Lebanon, Macao, Nepal, Oman, Pakistan, Philippines, Saudi Arabia, Singapore, Sri Lanka, Syria, T a i w a n , Republic of Vietnam UNITED STATES and CANADA UNITED STATES New York, N.Y. 5. LM Ericsson T e l e c o m m u n i c a tions, INC 2. Ericsson Centrum, Inc. CANADA Montreal 2. L M Ericsson Ltd. AUSTRALIA and OCEANIA Melbourne 1. L M Ericsson Pty. Ltd. 1. A.E.E. Capacitors Pty. Ltd. 5. Teleric Pty. Ltd. Sydney 3. Conqueror Cables Pty. Ltd. Representatives In: New Caledonia, Matinique, New Zealand, Tahiti. 1. Sales company w i t h manufacturing 2. Sales and installation company 3. Associated sales c o m p a n y with manufacturing 4. Associated company w i t h sales and installation 5. Other company 6. Other associated company 7. Technical office TELEFONAKTIEBOLAGET LM ERICSSON LJUNGLOFS. STHLM 1976