*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-* -=[SD]=- Sepulchral Darkness -=[SD]=- presents : WHAT IS SYSTEM X ANYWAY? Brought to you by : Keltic Phr0st Sepulchral Darkness '95 All Rights Worth Shit *-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-* Compiled by Aztech from Electronics, The Maplin Magazine In my opinion, this is an excellent article, concise and well written.Its inclusion in this collection should provide an excellent starting point for the ambitious phreak, as well as clarifying a few 'fuzzy' points for the more experienced. If I can obtain access to a colour digitiser, I will also include the accompanying colour photographs as .GIF's,but dont hold your breath... ***** ***** ***** ***** ***** ***** ***** ***** The Digital Exchange - How it works I nearly called this article "The digital telephone exchange" - out of habit and familiarity I suppose. But of course the whole point of the new network is that it has been designed to carry voice and data, narrow band and (eventually) wide-band, and is more properly described as a communications network. New technology is again blurring the differences, and telephones are just one type of terminal which can access the network. The previous article described the need for a a national digital comms network, its evolution, design principles and facilities. This article concentrates on how the modular, software controlled, family of digital switches actually fit together and what they do. One of the features of System X is its use of common channel signalling to replace the old analogue frequency division multiplexing. This may need explaining, so I'll do it now and get it out of the way. If you already know about it, skip a few paragraphs. Common Channel signalling : A small diversion --------------------------------------------- The use of common channel signalling enables much faster set-up times, and means that, during the course of a call, facilities such as call transfer, call diversion, and all the rest, can be invoked. Common channel signalling (CCS) meets the international standard known as CCITT #7. In CCS, Information about the caller,the called, call process and metering, is defined in a unique digital message. A common channel is dedicated to sending this message, together with messages defining other calls, thus providing signalling information for many seperate data/voice channels. CCITT #7 defines the different levels of signalling: o Level 1 specifies the bit rate (64 kbps), and the signalling information is bunched into messages, seperated by flags. o Level 2 adds error detection, requesting re-transmission if necessary. o Level 3 adds codes which define the origin and destination of the call,plus the intermediate switch nodes along the route. The route is 'chosen' from a memory based table, and takes account of any network problems. This level also provides system management information about the moment status of this part of the network. o Level 4 and up carry the information needed to connect you to your (typo, final word missing, possibly 'destination'?) As well as meeting the CCITT #7 protocol, System X also copes with other protocols already in use, such as those covering packet switching and private networks. Figure 1(a) and 1(b) illustrate the simplification available when using digital rather than analogue transmission and switching. Figure 1a and 1b. Comparison of analogue and digital transmission and switching. Signalling /----------\ +--------------+ +---- ----+--------+ | | +---- ----+ | | / \ | +---- One ----+ Multi -| Frequency Division ---->--------o o----->+---- per ----+ Plex |--------------------> | Electro - | +---- circuit ----+ | multiplex bearer | Mechanical | +---- ----+ | +--------------+ +---- ----+--------+ Group Switching Centre (a) +-----------+ | Digital | 2048 Kbps Time-Division | Switch |------------------------------------------------------> +-----------+ Digital Path multiplex bearer | | ^ | | <-------+ | | | | | +-----------+ | | | Control | | +---------------------------+ | Processor | | | +-----------+ +--Inter - Processor | Main signalling 30 Traffic paths Switching 1 signalling path Unit 1 alarms/sync/etc path (b) Back to the main theme: ----------------------- By the late 1970's, modernisation of the U.K.s telecomms network was long overdue. The existing network had been based on pre-war technology, and piecemeal upgrading was no longer enough. A radical re-think was required. Digital transmission of both speech and data signals had become possible on a large scale, and reliable enough to be considered as the basis of the replacement comms network. A range of differently sized exchanges was planned, using identical hardware subassemblies which could be plugged together like lego bricks to suit local requirements, and controlled by variable software modules. Figure 2 illustrates this network structure, with concentrators(C) interfacing telephones, faxes, telexes, computer modems(M), to the local exchange(L). Operators(O) are now connected at trunk exchanges, though they may be geographically seperated from the actual exchange equipment by many miles. Figure 2 : System X Network structure. +------+---> | | Trunks +---+ +---+ +-----> | O |--------| T | +---+ +---+ | | (Unknown) +---+ | | +---+ | +-| L |------------+ +------------| L |----+ | +---+ +---+ | | | | | | (C) | | +---+ (C) (C)------| M | +---+ The different sizes of exchange were planned to have the capacities shown in Table 1. The first set of numbers are those originally envisaged;the numbers in brackets are those actually available to the UK network, according to GPT's most recent publications. Some of the changes are remarkable. Table 1 : Traffic capacity Exchange Status Max Max Max lines/ switched call attempts circuits erlangs* per hour Large International 85,000 20,000 400,000 Medium International 8,000 2,000 50,000 Large Trunk 85,000 20,000 500,000 (60,000) (25,000) (1,000,000) Large Local 60,000 10,000 500,000 (100,000) (25,000) (800,000) Medium Local 10,000 2,000 80,000 Small local 2,000 160 8,000 (5,000) 1,000 (20,000) Concentrator 2,000 160 8,000 (6,500) (700) (22,000) * An erlang is defined as one line busy for one hour. It could also be two lines busy for half-an-hour each, or ten lines busy for six minutes each (but that would be a bit inefficient). Since the whole philosphy of System X was that it would be modular, the actual size of each exchange is designed to suit local requirements. For convenience in these articles, I have used the name System X (The British made system supplied by GPT) as a synonym for al the UK's digital exchanges, although a number of manafacturers are now involved. Photos 1,2 and 3 show what the exchanges look like - photos 1 and 2 are photographs taken from the front of the suites, and photo 3 shows what the back looks like without its tidy casing. But how does the digital exchange actually work, what changes have been made to enable you access to the national digital network, and what is happening to the existing exchanges? We visted our local exchanges to find out. The Subscriber Concentrator : Small but beautiful ------------------------------------------------- Concentrators exist at the very far flung ends of the national telecomms system. They are the interface between the subscriber and the network, and provide terminations for all kinds of domestic or buisness telecomms and data equipment. They are usually connected by copper or fiber cables to the local exchange, but it is possible to have concentrators so remote that they are linked to the 'local' exchange via geostationary satellites! It doesnt matter if your equipment is analogue or digital, or if you rent it from the local newsagent. If it carries that little green circle, the concentrator will process calls in and out, and cope with tone or pulse dialling, earth or loop calling, private metering, night busy switching; The works. The concentrator is also used by outstationed analogue or digital multiplexers for access to the main network. Our small town qualifies for a set of concentrators. The System X suites occupy less than 20% of the space use by the previous strowger exchange. Now that the old racks have been stripped out (some of them being sold for re-use, the rest being scrapped) the original building is being used as a maintennance depot and store for the local BT vehicle fleet. Each concentrator can output to a maximum of 240 lines connected to our nearest large town, where all the remaining call processing takes place. This is a good example of how the new network streamlines processing capability, and explains why our STD code changed too. The old autonomous local electro-mechanical exchanges, with their expensive and dedicated hardware, have been replaced by a much more flexible system. Calls from rural subscribers in a much larger geographical area are now channelled through local concentrators into the centralised facilities provided maybe 10-20 miles away. Of course, this means that my telephone call to a neighbour 200 yds away actually travels about 20 miles to get there. I could walk, but its raining. Our local exchange channels about 12,500 lines into six concentrators. The neccesary electronics are housed in four suites, each made up of six cabinets. Each cabinet is about 7ft high x 30in. wide x21in deep, and contains six shelves. Two of the shelves carry the subscriber private metering and other 'housekeeping' equipment, and subscriber linecards occupy the other four. The linecards themselves contain very few components - 3 DIL relays, about two dozen custom ICs, a few resistors, gas-discharge protection - on a card which measures 12ft. x 10in. x1in. There are three groups of line cards on each shelf; each group contains eight cards. Each line contains all the electronics to handle eight subscriber lines and provides automatic gain setting, line balancing, and the A-to-D conversion via codecs. Sorry about all the numbers, but they show how much physical space is being saved by the installation of these digital exchanges. 8(subscribers) x 8(Linecards) x3(groups) = 192 subscribers per shelf. That is, 192 subscriber terminations now occupy a space approximately 30in x 12in x 10in, which is about the same as a small bookshelf containg 40 Maplins catalogues. The equivalent activity (excluding any A-to-D conversion, of course) in an elctro mechanical exchange occupied a space of about 160in x 242in x 6in, which was more than six times bigger. The chap in photo four is dismantling an old uniselector, and we will not see the like of it again. To put its size in perspective, all the equipment needed for a 5,000 line small local digital exchange can be delivered in one 30ft container. Subscriber and network cabling ------------------------------ The links between our local exchange and the next one up the line are provided by an equal mixture of (existing) copper - carrying 30 PCM channels - and (newly installed) optical fibre. (Did you know that every copper cable under the ground is enclosed in an airtight sheath and compressed air is pumped through to keep the water out if there is a break in the sheath? I didnt, until I saw the gauges in the exchange. The cable air pressure is monitored constantly and any reduction in pressure will eventually result in a road being dug up to repair the damage!) Copper pairs are difficult and expensive to maintain, and the bandwidth is limited. Although it is claimed that as many as 10 digital voice channels can be carried on a single pair, most PTT's assume that the ISDN bit rate of 144 Kbps is a more practical limit. And although copper cable has a long life, it is prone to faults, signal loss, and water, and existing links are being replaced by optical fibre. In November 1985, it was expected that by 1988 40% of trunk traffic would be carried over fibre-optic lines, with promised major cuts in phone charges. Has anyone noticed a price reduction? Cost reductions were supposed to result from fewer maintennance staff being needed, and by the downward trend in the costs of fibre. The cost of fibre-optic lines has dropped substantially over the past few years, and now is near the cost-per-kilometre of copper lines. BT has already started laying fibre directly to large buisness customers (say above 25 lines), in an overlay network of 96-fibre cables, split into 48- or 24-cable links to a joint common to several customers. This is not economic for smaller subscribers, although BT is now experimenting with fibre in the local loop which connects clients premises to local exchanges. Such cabling would be capable of carrying voice, data, stereo television and radio traffic, though at present BT is not normally allowed to carry television channels. BT has announced that it will be spending œ5 million on a trial installation in Bishops Stortford, although a similar experiment in France has been quietly allowed to die because the local subscribers were not making enough use of the facility to make the installation economical. Power ----- Each suite of exchange cabinets contains its own backup sealed batteries and rectifiers. As a suite is added to extend the exchanges capability, its backup power is also automatically included. These batteries need no regular maintennance. Gone are the days of seperate large battery rooms in each exchange, with their memorable instructions (according to an engineer of my acquaintance) "All water used to top up these batteries must have been passed by the Assistant engineer". Three phase mains comes straight into each rack. The batteries provide one hour of support; a longer power outage is covered by a local diesel generator. The system architecture beyond the local concentrators ------------------------------------------------------ Once your call has passed through a local concentrator, it will enter the digital network proper. All subsequent call switching and transmission goes through a series of whichever larger and larger installations are neccesary to route your call (eventually) your destinations concentrator, and hence to your mum. Modular assembly results in the different sizes of exchange already mentioned and shown in table 1. A small local exchange (up to 5,000 lines) can support small concentrators and multiplexers. Although it is designed to operate unattended, it can offer full operator facilities based at the trunk exchange, and is monitored remotely for traffic statistics, fault analysis, and other management information collected further up the line. Larger exchanges are (nearly) just more of the same,plugged together. So what are these modules? Figure 3 shows the switching subsystem architecture and these modules, plus the over-riding control systems, are described next. Figure 3 : System X switching subsystem architecture +-------------------------------------------+ +---+ | | | | | +-----+ +-----+ +-----+ +-----+ +-----+ | / |-----| OSS | | MSS | | MCS | | OCS | | CPS | | +----+ | +-----+ +-----+ +-----+ +-----+ +-----+ | Human | +-----+ +-----+ +-----+ | Access | | MTS | | DSS | | SIS | | | +-----+ +-----+ +-----+ | | | | | PS | +-------------|-------|-------|-------------+ | | | To and from | | | To and from local exchanges +-----+ +-----+ +-----+ trunk exchanges in | MTS |-| |-| SIS | in +-----+ | | +-----+ | | +-----+ +------+ | | +------+ +-----+ analogue { -------| SIS | ---| ALTS |-| |-| ALTS |---| SIS |---------- } EN{ +-----+ +------+ | DSS | +------+ +-----+ } EN { | | | | | | } { +----------+ | | +----------+ digital } { ---------------------------| |----------------------------- } | | digital N.N. ---------------------------| |----+ +---------------------- N.N. +-----+ | | | +---+ +------|NSS| +---+ EN = Existing Network NN = New Network The Hardware ------------ The controlling heart of the digital switch is the PROCESSOR. Because of System X's inherent ISDN capability, the designers knew that there would be a great increase in the number of call attempts caused by the inclusion of data transmissions. Unlike voice links, data can be sent in very frequent, very short bursts. But each data transmission is still a call, as far as the switch is concerned, and has to be handled just like a voice transmission. Each System X processor was therefor designed to cope with up to one million busy hour call attempts (BHCAs) and more if needed. It achieves this processing power by using standard components in an un-usual architecture, in which up to four central processing units are close coupled, sharing common memory. These clusters, as they are called, can then be loose coupled in a group of up to eight clusters - such a switch size would handle 2 million BHCAs. The CENTRAL PROCESSOR UNIT CLUSTER Processes all the information needed to pass a call through the exchange. It introduces great flexibilty in handling;for instance it can dictate changes to line allocations, so that known heavily used lines are allocated greater access to the concentrator. Always under the contro of an operating system, the CPU runs background maintennance and diagnostic checks in between processing calls. In the event of a CPU failure, the remaining CPUs in the cluster can maintain service because they have common memory access. The DIGITAL SWITCH connects all the sub-systems. Its main function is to switch 64Kbps paths, although it is also capable of switching up to 34 Mbps. It handles common-channel signallin, and interworking signalling from the remaining analogue exchanges. All signal path setups are decided by an internal microprocessor and are then duplicated within the switch. This ensures continuity of service in the event of path failure. The PROCESSOR SUBSYSTEM (PS in Figure 3) decides the operation of the various sub-systems, using software programs to control call processing, system overload and maintennance, and generates management and operations statistics. The SUBSCRIBER SWITCHING SUBSYSTEM (SSS) accepts and analyses incoming calls, and concentrates them into high-traffic common circuits at the local exchange. The DIGITAL SWITCHING SUBSYSTEM (DSS) interconnects these circuits. The MESSAGE TRANSMISSION SUBSYSTEM (MTS) carries out common channel signalling functions, and provides error correction. The modules use normally two, sometimes four, signalling links, routed via a spare timeslot (usually TS16). In heavy traffic signalling messages are repeated to ensure their transmission, and the whole module operates within a choice of three hardware routes, to minimise loss or failure. In light traffic, the unit transmits signals which allow synchronization to be checked and gives local status information. Analogue signalling links terminate here,via modems. Not only are outgoing and incoming messages queued here, but message information is stored in case any internal signalling link fails. In this case, the information is re-transmitted over one of the the alternative internal links. Voice transmission, by its nature, conatains redundancies. Synchronization is not as critical, since some information can be lost without the (human) receiver noticing any difference. But synchronization is essential in a network which carries data, since loss of even one piece of information could turn the message into garbage. To avoid this, one of the modules is responsible for keeping the exchange 2.048 Mhz master clock in synchronization first with other master clocks and ultimately with the reference clock for the whole network. As a back-up, System X provides buffer stores at the receiving end of each link, so a small amount of drift, either phase or frequency is catered for. The SIGNALLING INTERWORKING SUBSYSTEM (SIS) interfaces the exchange to the different signalling systems in other types of exchange, with the exception of MF, which is terminated seperately but in a similar way.This Subsystem also provides recorded announcements and standard tones. Incoming signals are converted to TS16 format for analysis and processing, and up to 900 circuits can be handled by each TS16 unit. The ANALOGUE LINE TERMINATING SUBSYSTEM (ALTS) converts analogue transmission signals into a digital format, and digital to analogue, to allow existing analogue exchanges to be used until they are replaced by digital systems. The NETWORK SYNCHRONIZATION SUBSYSTEM (NSS) ensures that the exchange operates within the binary rate tolerance of the network. The software ------------ The software used in the network is also modular, under the control of a central operating system which allocates time and storage to the various applications modules, and acts as timekeeper and messenger. The CENTRAL PROCESSOR UNIT controls the working of all the software subsystem processes on the basis of priority.Each process can be in one of five states: RUNNING (being carried out by the CPU) SUSPENDED/INTERRUPTED (paused, for higher priority processing) SUSPENDED/UNBLOCKED (waiting for space on the CPU) BLOCKED (process has finished) TRAPPED (fault prevents further operation) Thre CPU deals with one process at a time. At any time it can be instructed to suspend and store - 'nest' - the activity because a higher priority process is ready to be dealt with. By constantly monitoring the priority of suspended processes, and dealing with the highest first, the CPU is designed to handle all the subsystems with efficiency and maximum throughput. And, unlike humans operating under these conditions, the CPU does actually get those lowest priority jobs done! The CALL PROCESSING SUBSYSTEM (CPS) deals with network routing, route selection, route supervision, numbering, control of user facilities. The CPS receives and stores the incoming message (from the MTS) earmarks an area of memory for a call supervision record, and requests the DSS to connect the incoming circuit to an outgoing one. When the circuit is completed, the message is erased, although the call supervision continues until the circuit is cleared. The OVERLOAD CONTROL SUBSYSTEM (OCS) monitors processor activity. Abnormal overloads could cause failure in the central processor, or problems in the network. If the number of call attempts is more than the processor can handle, newly arriving calls can be rejected instead of being allowed to clog up the queues. This means that the number of successful calls -those already inthe queue-is maintained at an optimum level, and a greater throughput is acheived. Succesful connections earn money; rejected callers will usually try again. The MANAGEMENT STATISTICS SUBSYSTEM (MMS) interfaces with other subsystems to collect information for later analysis on such useful parameters as traffic intensity, route usage, call failure, calls in different tariff bands, and local congestion The MAINTENNANCE CONTROL SUBSYSTEM (MCS) monitors the systems operation, identifies actual or possible faults, isolates faulty sections and generates reports for maintennance staff. A number of automatic tests can be generated, including checking subscriber lines and exchange interconnections. The OPERATOR SUBSYSTEM (OSS) enables operators and maintennance staff to communicate with the system, and supports printers and and data links for the system to 'talk' back. Even System X still needs human beings occasionally. Operators now work in normal offices, and they can be some distances away from the main exchange. After all, why pay for prime city-centre office space when the technology allows them to be out-stationed, maybe even with a view of trees rather then traffic? Their work is controlled and presented by an operator services subsystem, which allocates priorities to all incoming calls. Each operator has a monitor, a keyboard, and a headset, and full details of the next call are presented on-screen (and to the headset) by the operator services subsystem.The operator initially has access to al the facilities likely to be required by the caller, and the software is designed to make processing a call as quick and effecient as possible. The receipt of an emeregency call, for instance, will produce a list of all the local emergency service names and telephone numbers; the operator simply selects the correct service from the menu and the system takes over the job of connecting the caller. Processing directory enquiries has also been automated to a greater extent than before. The operator has access to the database of telephone numbers; once the required number is located on the screen, the tap of a key frees the operator to take the next call, and invokes a recorder voice which clearly and patiently tells us the number we want. Other databases exist. The opertor can enter an exchange name, and the system will translate this into its dialling code. Freephone names are similarly translated automatically. And if your telephone credit card is invalid, the operator will soon find out. Network upgrading ----------------- The modular structure enables enhancement of specific subsystems without disturbing the rest of the network. Since so much of the network is software controlled, system upgrades tend to be in the software rather than the hardware. Upgrades are already being installed. The most noticeable improvement has been in the reduced time taken to restore service after a CPU crash. Early systems had to re-program each concentrator in sequence - based on PROMs, re-programming each concentrator took about ten minutes and it could be hours before all the concentrators controlled by the CPU were ready. Newer systems re-program all the concentrators in paralell and the local network can be up and running again in about 30 minutes. CPU crashes do happen, although the stability inherent in the design more than meets the national standards. System Maintennance ------------------- Test end maintennance is monitored and controlled at regional centres (ours is done via a VAX in Bristol) with access via a terminal in each local exchange. So access to 'our' CPU at Gloucester is via a data link to the VAX in Bristol, and a further link back to Gloucester. In the old electro mechanical exchanges, the worst that could happen was a total exchange power failure, which was almost unheard of. The weakness of the new network is that a software fault could, in the worst case, affect subscribers in a much larger catchement area. In practise however, an accumulation of faults should result in a controlled reduction of quality of service, rather than a system shutdown. Our local BT maintennance engineer was however pleased with with the reliability of our exchange. It has a very low failure rate, and he had only made four maintennance visits in the three monthes before our visit. The exchange 'spares' - three spare line cards and one of everything else - are kept in one cupboard! So, is digital desirable? ------------------------- It certainly seems to be user freindly, flexible and reliable. Many of the items on my personal wish-list for communicating with the outside world are nowavailable. But there is still one at the very top of the list - when will an ordinary domestic telephone tell me who is calling BEFORE I pick up the handset? ***** ***** ***** ***** ***** ***** ***** ***** Its not exactly hot off the news stand, so one or two things have already been and done. For example, The TV on demand experiment and Caller ID. But 99% of this file is accurate, and should clarify some grey areas in your knowledge of the British PSTN. I found this article a boon, and will be looking for more like it in the future. ---=[AZTECH]=---