Reference / Paper · 1979
Telecommunications Switching Principles
Modern switching theory text covering step-by-step, crossbar and stored program control.
Reference / Paper · 1979
Modern switching theory text covering step-by-step, crossbar and stored program control.
Telecommunications Switching Principles M. T. HILLS The MIT Press Cambridge, Massachusetts, and London, England Second printing, 1982 First MIT Press edition, 1979 First published in England in 1979 by George Allen &. Unwin Ltd. @ 1979 M. T. Hills All rights reserved. No part of this book may be reproduced in any form or by any means, electronic or mechanical, including photocopying, recording, or by any information storage and retrieval system, without permission in writing from the publisher. Printed in the United States of America library of Congress catalog Gard number: 78-61815 ISBN 0-262'()8092·3 TELECOMMUNICATIONS SWITCHING PRINCIPLES Dedicated to A. Strowger the originator of the world's first and largest real-time multi-processor control system Preface Telecommunications networks for public and private use are possibly among the largest interconnected systems in the world. The world-wide public telephone service is provided by a network interconnecting more than 3 50 million tele phones. A telecommunications network consists of terminals, transmission links and switching centres. The principles underlying the design of the transmission and terminal components are well established and have a coherence. In contrast there appears to be a multitude of different approaches to the design of the switching centre component. One object of this book is to develop a coherent set of principles for the design of these centres. Several practical examples are described of different systems around the world to demonstrate the application of the principles and to give the reader a background knowledge of the state of art. Most of these examples are taken from the public telephone network as it is in this network the principles were evolved. However, the principles are shown to apply equally to other types of switching systems such as data networks or private telephone systems. The original intention was to include a wide range of system descriptions but reasons of economy dictated their omission. The choice of systems actually included was random and omissions include many of the major systems currently available. The majority of switching equipment in public use today is still built from electro-mechanical components such as relays. Much of this technology will be unfamiliar to recently qualified engineers and some of the book is devoted t o describing its application. The use of computers and, more recently, micro-processors is bringing a revolution in the design of switching systems, and a significant proportion of the book is devoted to such systems. However, it is suggested (in Chapter 13) that some of the earlier design of computer-controlled Switching centres could have been improved if they had more fully adopted the design principles developed for the electro-mechanical systems. The book should be of value to an engineer with an electronic engineering, computer systems or conventional switching background who has either to design or to operate a switching system or who needs to use one. It could form the basis of a final year course in telecommunication engineering, teleprocessing or real-time computer systems. Apart from some mathematics associated with the traffic and queueing theory chapter, the only prerequisite knowledge is that of basic electronics and in certain sections a background knowledge of com puters and programming. x PREFACE The book is based on courses given at postgraduate level in the University of Essex M.Sc. in Telecommunications Systems, to numerous classes of new entrants to the British Post Office and to many audiences around the world. The book in its current form has been made possible by the interaction with numerous colleagues within the Electrical Engineering Department, University of Essex and in the British Post Office, G E C , Plessey, Standard Telephone and Cables, Pye-TM C and Thorn- Ericsson in the U.K. Abroad, the strongest influence has been from the Nippon Telegraph and Telephone Public Corporation (NIT) where I spent a sabbatical leave in their Electrical Communications Laboratories in the Tokyo ar'O',a. I am particularly grateful to Malcolm Hamer who has painstakingly read several drafts of the book and in many detailed discussions assisted me in removing much of the obscurity in the earlier drafts. I should like to express my gratitude to the British Post Office whose financial support of the Telecommunications group within the Department of Electrical Engineering Science has made possible the environment in which the concepts described in the book have been developed and tested. My hope is that some of the ideas discussed in the book will go towards improving the design of telephone switching systems. Finally, I must thank my publisher Roger Jones for his apparently groundless faith that I would actually get the book finished albeit five years later than originally planned. Without his continued encouragement even this version would not have appeared. Contents page J Introduction to switching systems 1.1 1.2 1.3 1.4 1. 5 1.6 2 Telephone systems Data and message switching systems Switching elements 27 32 36 Introduction Functi onal subdivision Comm on switch netw orks and their control Control of switching systems Alternative organisations o f switching netw orks Time-shared decisi on making Stored pr ogram c ontrol 54 55 62 64 72 76 83 Traffic theory 4.1 4.2 4.3 4.4 4. 5 4.6 4. 7 5 3 4 12 18 21 The design of economic switching centres 3.1 3.2 3.3 3.4 3. 5 3.6 3. 7 4 I Signalling and switching techniques 2.1 2.2 2.3 3 Purpose of the bo ok Types of switching systems Centralised switching systems Basic switching centre m odel Resource sharing Introduction to traffic and queueing theory Basic equations Queueing systems C omparison of the various distributi ons Applicati ons of traffic theory Multi-stage switch netw orks - conditional selection Simulati on techniques Traffic measurement and traffic prediction 86 92 93 97 IDS 108 I I1 Telephone network organisation 5.1 5.2 5.3 5.4 5. 5 5 6 5.7 . Network planning Routing plans Numbering plans Register c ontrol of networks Use of step-by-step equipment Charging Network management 113 114 129 1 32 1 34 137 1 38 xii 6 CONTENTS Practical signalling systems 61 6.2 . 7 9 Basic multi-stage networks Use of mixing stages Network and channel graphs Networks with concentration Lee's simulation technique for evaluating blocking probabilities Role of control units 8.2 The arbiter 8.3 8.4 State transition diagram for a called control unit Signalling between control units 8.5 Signalling techniques 8.6 Design of a control unit for two-way traffic Some circuit techniques 9.1 Basic elements Electronic components 9.3 9. 4 9. 5 Realisation of arbiters 9.6 Third wire control Junctors Switch matrix control 10.4 Reed relay systems 195 198 205 213 21 5 218 222 226 2 32 232 239 2 50 Computer controlled switching systems 11.1 Software organisation of computer controlled centres 11.2 No. 1 ESS 11.3 Japanese D-lO 11.4 Metaconta systems 12 181 182 184 19 3 Practical examples of switching systems 10.1 Introduction 10.2 Step-by-step 10.3 Crossbar systems 11 160 1 67 1 70 175 17 9 Control unit design 8.1 9-2 10 143 147 Design of switching networks 7.1 7.2 7.3 7.4 7.5 8 Types of signal for telephone systems Signalling techniques 258 267 277 284 Digital switching systems 12.1 Time-division switches 287 293 12.2 Relative merits of digital and analogue switching 12.3 Practical application o f digital telephone system - Bell No.4 ESS 2 9 5 CONTENTS 13. The future - xiii a personal opinion 13 .1 A critique of switching system design 13.2 Design principles for switching centre design 13.3 In summary 296 299 301 Appendices A. Optimal size of switching machine B. Deriv:ltion of some basic traffic theory results C. Lower limit on number of crosspoints needed in multi-stage networks D. Simplified proof of Takagi optimum channel graph theorem E. Estimation of the traffic capacity of two-stage group selectors 303 306 311 314 322 Tables 1.1 Names of different level switching centres in telephone networks 1.2 Some typical availability objectives for public telephone systems 2.1 Example of telex signalling 10 2.2 Example of F31 telegram message format 34 4.1 Some results for a two-server system 4.2 Results for two-server system with infinite queue 94 94 4.3 Relationship between offered traffic level and server efficiency 99 12 33 4.4 Blocking probabilities of networks shown in Figure 4.12 108 5.1 Comparison of mesh and star connections 5.2 Calculation of alternative routing strategy 116 6.1 Rl signalling system (N. American system ) 6.2 R2 signalling system 121 152 153 7.1 Blocking probabilities for simple multi-stage networks 167 7.2 Comparison of networks of Figure 7.12 179 11.1 Causes of down time in No. 1 ESS D.1 Values of F (i,i) corresponding to different states of the two E links 27 5 D.2 Values of Ba and Bb for a = � 317 3 19 1 Introduction to switching systems 1.1 Purpose of the book The world-wide public telephone system is remarkable by virtue of the wide variety of equipment used in its construction from mechanical devices installed in the late 1920s to the most modern miniaturised digital circuitry. It all has to work together to provide an economic and reliable service. Although it is the most visible switching system, the telephone system is not the only one. Other examples of switching systems are the public telex network allowing dial-up communication between teleprinters, and a range of national and international networks for company, bank and military users. Of increasing importance during the 1980s will be the growth of data switching which many people predict will overtake voice communication (in volume) by the end of the century. There are other systems which at first sight may not be regarded as switching systems but whose design, as this book shows, is based on the same principles. These include telemetry systems, on-line access to central computers from widely dispersed devices such as visual display terminals, card readers and so on. The purpose of a telecommunications switching system is to provide the means to pass information from any terminal device to any other terminal device selected by the originator. Three components are necessary for such systems: terminals which are either input or output transducers. They convert the information into an electrical signal at the transmitting end and convert th.e electrical signal back into a usable form at the receiving end. A further function of a terminal is to generate and transmit control signals to indicate the required destination of the information signal. transmission link s to convey the information and control signals between the terminals and switching centres. switching centres to receive the control signals and to forward or connect the information signals. 2 INTRODUCTION TO SWITCHING SYSTEMS Transmission links are covered in detail in a companion volume [I] and receive only scant mention in this book. Terminals are covered only from the point of view of their capabilities to generate and receive the control signals. The book deals with the design of the individual switching centres and their incorporation in switching networks. The organisation of this book has two aims: (l) To show that there is a unified set of principles behind the wide range of superficially different switching centre designs, and to show how these principles may be applied to switching centre design using modern tech nology. To give a description of the implementation of the design principles in (2) some of the switching centre types in use today. This chapter introduces the basic systems concepts and the objectives of a system design. Chapter 2 describes the basic signalling and switching technique used in voice and data switching centres. The design of economic switching centres is covered in Chapter 3. Economic design involves resource sharing and resource sharing implies that there is a probability that a resource (such as a switch or a transmission link) will not be available at the instant it is required. The design of a switching centre therefore involves determining the number of resources required to achieve a particular probability of no resource being available (and possibly a particular length of wait for a resource to become available). This is the subject of traffic theory which is covered in Chapter 4. Switching centres are organised in networks and these are discussed in Chapter 5. The practical means by which control signals are passed from centre to centre are the subject of Chapter 6. The next two chapters are more theoretical and in Chapter 7 a coherent approach is given to the design of switch networks. Of particular importance in this chap.ter is a discussion of the work of Takagi on optimum channel graphs. It is thought that this is the first time that this treatment has been covered in a text book. Chapter 8 is a theoretical approach to the design of control systems and attempts to show a unified approach to the design of electronic and computer controlled systems and the relationship to signalling systems. After the theoretical chapters, the remainder of the book deals with practical aspects of telephone Switching systems. Chapter 9 describes a wide range of the practical techniques used within electro-mechanical and electronic systems. A selection of electro-mechanical and electronic systems are described in Chapter 10 and the examples are carefully chosen to demonstrate the application of particular prinCiples. Computer controlled systems have Chapter II to them selves and some of this chapter requires a basic knowledge of computers. The long-term future for the technology of switching systems is almost certainly going to be digital and the basic differences to other systems together with an example are discussed in Chapter 12. TYPES OF SWITCHING SYSTEMS 3 Finally, the author has allowed himself some licence and the last chapter is in the form of an editorial, which attempts to predict the future direction of system architectures. This chapter attempts to show that the centralised control systems origins come from the 1960s and are not the way today's systems should be-designed. The centralised systems appear to have been designed as an exercise in programming rather than as the design of a reliable telephone switching system using principles described in this book. Time will show who was right. 1.2 Types of switching systems A system similar to (but smaller than) the telephone network is the public telex network (telegraph exchange) which provides world·wide direct interconnection of teleprinters. Both the telephone and telex networks are examples of what are called circuit switching since they set up a circuit between two terminals which then interchange information directly. Another class of system which is more familiar to the business or military user is that of message switching. The terminals of message switched systems are usually teleprinters, but unlike the telex network they are not interconnected directly. Instead, when a terminal user types a message destined for some other terminal, the system stores the message and delivers it to the required terminal at some later time. The reason for the delay is that the system is designed to maximise the utilisation of transmission links by queueing messages awaiting the use of a link. In order to set up a direct connection over many links connected end-to-end it is necessary for each link to be simultaneously free. As will be seen later, this implies that the average utilisation of the links must be low if the probability of a direct connection being available on demand is to be high enough to satisfy most users. However, in a message switched system, where messages are queued for each link, a much higher link utilisation is achieved. Another name for this type of system is store and forward switching. The advent of real-time computer systems for airline reservation, banking systems and remote data processing in general has been based on the use of telecommunications networks to carry data between computer-type terminal devices and large real-time computers. Such applications may be served by a purpose-built circuit switched network or by another system such as the public telephone network in conjunction with special signal processing techniques. More recently, general purpose packet switching systems have been developed which take the data from a terminal or a computer and transmit it as short packets of information to the required destination. Such systems are midway between the two extremes of circuit switching and store and forward. The interchange of packets may be made so rapid that a terminal appears to provide a 'conversational' connection while at the same time high transmission link utilisations are obtained through queueing. 4 INTRODUCTION TO SWITCHING SYSTEMS 1.3 Centralised switchin g systems A simple way of s truc turing a swi tched network is to arrange tha t each terminal has a direc t transmission link to every o ther terminal, as shown in Figure 1. 1a. Each terminal needs a swi tch to connect i t to the required link and a swi tch to make connection to a link in order to receive an incoming call. For N term inals this arrangemen t needs a to tal of %N(N - I) links. (al Figure 1.1 (a) Full interconnection for five terminals. Some form of switching is needed within each terminal to connect it to the appropriate link. (b) Use of one channel per terminal. Each terminal is permanently connected to one channel and all other terminals may access a particular terminal by operating a switch which connects it to the appropriate channel. (c) As in (b) but with less than N channels. An alterna tive approach which needs only N links is to provide one link per terminal and to arrange tha t all other terminals have access to i t as shown in Figure 1. 1 b. This simplifies the terminal equipment because it removes the need to connect a terminal to a link for an incoming call. Th is is a practical arrangement for systems such as house telephones or intercoms where there is a relatively small number of terminals close toge ther. For ins tance, it is possible to provide an eleven·terminal system with each terminal havin g ten bu t tons and eleven pairs of wires going around to each instrumen t. However, when the number of terminals increases, or their geographical separation increases, the cos t of cabling makes this arrangemen t uneconomic. Another example of the technique illustrated in Figure 1. 1 b is a radio telephone ne twork in which each terminal is given i ts own frequency. An origina tor can set-up a call by tunin g his transmi t ter to the frequency of the called terminal. CENTRA LISED SWITCHING SYSTEMS 5 (b) i------r-------------, order channel }� i11l ill! illl I I annels (c) It is possible to use an arrangement si milar to that just described but wi th fewer than N links, as shown in Figure 1. 1 c. Wi th this arrangement it is ne cessary for the calling terminal to select a free link and for the called terminal to be connected to the same link. One technique used extensively for radio telephones involves a com mon channel to which all free terminals are connected [2]. A calling signal (which may be coded or verbal) is sent along this link and is re ceived by the terminals. The calling signal infor ms the called terminal (or i ts user) of an in co ming call and indicates the link to whi ch the terminal should be switched in order to receive the call. The recognition of the calling signal and the switching operations may be performed auto matically in a system using coded signals. In simpler systems the users may be required to listen ou t for, and act upon, verbal calling signals. At the present ti me this technique of 'distributed' switching is applied only to s mall telephone systems and to some radio telephone networks. One of the possibilities of the future is what is called a 'digi tal r ing main' in which a high speed binary digital link is connected to a large number of terminals. Similar techniques to that described above may then be used. For example, a terminal could connec t itself to a free time slot w ithin the digi tal bit s tream in order to set up a co m muni ca tion path to the called terminal. 6 INTRODUCTION TO SWITCHING SYSTEMS -- information link - - control signals (a) Trunk links between switching machines. shared between all terminals as needed (b) Firure 1.2 The use of centralised switching machines. (a) Single machine whlch reduces average length of transmission link as compared to Figure 1.1. (b) Use of additional switching machine to reduce transmission costs further (if the terminals have a low utilisation). CENTRA LISED SWITCHING SYSTEMS 7 In mos t practical swi tched ne tworks i t is usually more economic to provide a link be tween each terminal and a central location Centralised switching centres. and to perform all swi tching operations there, as illus tra ted in Figure 1.2a. This arrangemen t significan tly reduces the to tal transmission cos ts in the ne twork. However, the swi tching cen tre mus t be operated by remo te con trol from the terminal and this tends to increase the to tal swi tching costs. The to tal transmission cos ts may be reduced even further if a number of local swi tching cen tres are used ins tead of one na tional cen tre because this reduces the average leng th of the connection be tween a terminal and i ts neares t swi tching cen tre ( Figure 1.2b). The local cen tres mus t be in terconnec ted by transmission links which are usually called trunks. These trunks are shared by all the terminals connec ted to each centre, and as will be shown later, the number of trunks con nec ted to a local cen tre can be very much smaller than the number of terminals. The use of mul tiplexing techniques for long distance transmission makes �os t per uni t dis tance for a trunk less than the cos t per uni t dis tance for the link be tween a terminal and i ts local switching centre. Therefore increasing the number of swi tching cen tres lowers the to tal transmission costs. However, as the number of centres is increased the to tal swi tching cos ts tend to increase for two reasons. Firs t, the local cen tres become more complex because they mus t be able to decide on a sui table routing to another cen tre and because the centres involved in a call mus t be able to exchange information. Secondly, economy of scale is los t wi th an increased number of local cen tres because two half-size centres, plus thei r buildings and power supplies, cos t more than one full-size centre. In general there is an op timum number of local cen tres for minimum to tal cos t of transmission and swi tching. This op timum number of local cen tres depends upon the relative cos ts of switching equipmen t and transmission equip men t and the geographical dis tribu tion of terminals. If cer t ain assump tions are made abou t the geographical dis tribution of terminals, their traffic charac teristics, and the cos ts of swi tching and trans mission, i t is possible to make ma thema tical analyses of the minimum cos t for a to tal sys tem [3] (see Appendix A). However, these analyses have limited value because the practical de tails of a given si tua tion invalida te the generalised assump tions. For ins tance, in the telephone field, de tailed cos tings of possible ne twork plans are needed in order to decide how many swi tchin g centres should be ins talled to cover a particular area , or, more usually, to decide whe ther i t is be t ter to ins tall a new switching cen tre in the suburbs of a growing town, or to extend the main cen tre [4]. These cos tings are simplified by using de tailed computer models of the area under consideration to facili ta te rapid es tima tion of the cos ts of al ternative arrangemen ts. Hierarchical systems. As the number of separate switching centres increases the number of differen t trunk rou tes be tween them increases. Above abou t ten cen tres the number of trunk rou tes becomes very large and rou tes tend to contain too 8 INTRODUCTION TO SWITCHING SYSTEMS @ G local switching centre trunk switching centre Figure 1.3 Mesh of trunk routes. few circuits to make the ne twork economic. The same argumen t tha t led to cen tralised local switching now applies to the trunk routes, so trunk swi tching centres are in troduced to swi tch be tween trunks. Figure 1.3 shows how the trunk centres reduce the total leng th of trunk circui ts. However, this arrange ment in troduces the addi tional cos t of trunk swi tching cen tres and also necess i ta tes three swi tching poin ts (rather than one or two) on some connec tions. The process of cen tral ising swi tching centres can occur at several levels leading to wha t is called a hierarchical network. This is best explained by reference to the public telephone ne twork. In a coun try there is normally a number of local swi tching centres, each serving anyth ing from 20 to 10 000 term inals (or even up to 100 000 in special cases). These local centres are ga thered in to groups and each group is served by a trunk cen tre wh ich can connec t calls be tween local cen tres within the group. For calls be tween terminals of one group of local centres, and those of another, the trunk centres the mselves have to be in terconnec ted by 'super' trunk centres each covering an area consis t ing of many local groups. Usually there is a number of these wider areas and these too will be in terconnected by higher level trunk centres. This principle is ex tended to the level where the number of trunk centres a t tha t level is small enough for i t to be prac tical to in terconnec t them fully wi th a mesh of trunk routes s imilar to those shown in Figure 1.3. The resul t is as shown in Figure 1.4a. CENTRALISED SWITCHING SYSTEMS ! / "..,- - ----- -- 9 , ( \ I \ (a) .... - � --- ....- o local centre D primary centre D secondary centre secondary centre trunk network L.---._-> (b ) local network { Figure 1.4 (a) Hierarchial network of switching centres. representa tion. primary centre local centre (b) Conventional 10 INTRODUCTION TO SWITCHING SYSTEMS Table 1.1 Names of different level switching centres in telephone networks (The numbers in brackets indicate approximate numbers in each country ( 1976).) Function International name U.K. name North American name Local Exchange Local exchange (6200) End office or Class 5 office (18 000) First level trunk Primary centre switching centre Group switching centre (370) Toll centre or Class 4 office (I 500) District swi tching centre (27) Primary centre or Class 3 office (250) Third level trunk Tertiary centre switching centre Main switching centre (9) Sectional centre of Class 2 office (65) Fourth level trunk Quaternary centre switching centre (not required in U.K.) Regional centre or Class I office ( 18 in U.S.A., 2 in Canada) Tandem exchange Tandem office Local switching centre directly connected to terminals Second level trunk switch ing centre Centre which provides con nections only between local centres (Le. no access to trunk network) Secondary centre Tandem exchange Switching centre which provides access to inter national net work. Centre du transit 3 (or CT3) International transit centre Centre du transit 2 (or CT2) Top level inter national transit switch ing centre (fully interconnected) Centre du transit I (or CT 1) (there are 7 of these in the international network) In the telephone system, differen t countries have adop ted different names for these different levels of centre. So to simplify discussion a se t of names has been in terna tionally agreed [5]. Table l. l shows these names toge ther wi th equiva len t U.K. and North American terms. Figure lAb shows a conventional way of illustra ting a hierarchical network plan. It can be seen from this that a hierarchical ne twork, as described above , CENTRALISED SWITCHING SYSTEMS 11 guarantees tha t a conne ction between any two terminals will be possible. Also it sets a limit to the number of links re quired in the wors t case. In pra ctice the situation is more complex. Large numbers of dire ct rou tes are provided between swit ches if the traffi c level is justi fied. So, for example, there may be dire ct routes between a tertiary centre in one area and a se condary centre in an area served by a different tertiary centre. Functions of telephone switching systems. Some of the functions of te lephone switching will now be defined. The loca l swit ching centre must rea ct to a calling signal from a terminal and must be able to receive informa tion to identify the re quired destination terminal. It must be able to de cide from the input informa tion whether the re quired terHlinal is conne cted to the same lo cal centre or whether a trunk conne ction is ne cessary via one or more intermediate trunk centres. If an intermediate trunk centre is needed the lo ca l centre mus t find a free trunk on the re quired trunk routes and conne ct the terminal to it. Further information must then be forwarded to the intermediate trunk centre or cen tres to progress the call to i ts des tination. Once a path has been set up from the originating centre to the terminating cen tre, the called terminal mus t be rung; and on ce the called party has answered, a spee ch path must be established between the two terminals for as long as the call lasts. Sin ce public te lephone systems must make money, at some s tage it is necessary to extra ct charging information for billing purposes. A further re quirement whi ch is not obv ious from what has been said so far is that a telephone syste m mus t be very reliable. In the language of reliability ma themati cs, a telephone system must have a high availability. Mos t SW i t ching systems are required to give uninterr upted servi ce for many years. Telephone sys tems, for example, have design lives of from 20 to 40 years_ Present te ch nology is su ch tha t no system can be guaranteed to be completely free of faul ts for this leng th of time, but i t is nevertheless possible to design a sys tem to provide an adequa te service even in the presen ce of faults or malfunctions. System reliability can be expressed ma thematically in terms of availability defined as: A= up-time up-time + down- time where the up-time is the to tal time tha t the system is operating sa tisfa ctorily and the down- time the total time that it is no t. An alternative and e quivalent definition of availability is in terms of the mean time be tween failures (m. t.b.f.) and the mean time to repair (m.t.t.r.): A= m.t.b.f. ---- m.t.b.f. + m.t.t.r. 12 INTRODUCTION TO SWITCHING SYSTEMS Table 1.2 Some typical availability objectives for public telephone systems For faults causing complete loss of service for more than 3 minutes and - m.t.b.f. ;;;. 10 years affecting only a single terminal - m.t.b.f.;;;" 20 years affecting 10% of terminals affecting complete switching centre - m.t.b.f.;;;" SO years Total down-time (due to switching centre failures) � 2 hours in 40 years i.e. overall availability;;;" 99'9994% Availability objectives are difficult to define. A fault which upsets the service of only one or a small number of users is less troublesome than a fault which makes a complete switching centre inoperative. The period of time that service is denied is also important. For telephone users, a break in service of only a few seconds (if it occurs only rarely) is not too troublesome, but breaks of fifteen minutes or more are very troublesome. Hence the availability figures must be split into a number of different categories. Some examples are shown in Table 1.2. It should be noted that the origin of these objectives is largely historical; early systems (step-by-step or Strowger switching systems to be described later) actually achieved availabilities of this order. 1.4 Basic switching centre model Most of the fundamental principles of switching system design can be under stood by considering a single centralised switching centre like that of Figure 1.2a. In addition to a channel for transfer of information, there is also a two·way path between each terminal and the centre for interchange of the system control signals. In fact, in most of the existing systems the same physical channel is used for both purposes. One of the practical problems of system design is the separation of the control signals from the information at the switching centre. The general realisation of this centralised machine (for circuit switching) is shown in Figure 1.5 where each terminal has its own switch and control signal path to its own control unit. Each switch has outlets to each of the other terminals. Also each control system has access to each of the other control units; this is necessary so that the control unit associated with a calling terminal may test whether the control unit associated with a called terminal is free and, if it is free, busy it. Signal exchange diagrams. The first step in any system design is to consider the range of control signals that has to be interchanged between a terminal and the system. This information is conveyed in the form of signals and many different ways are used to code these Signals. In telephone systems a commonly used form of signal is the changing of the value of an analogue quantity, such as 13 BASIC SWITCHING CENTRE MODEL r-------------------i to terminal 1 I I I I 0 ./ �-'� I - I I ___ I I I I I L � I 0 II · control unit 1 ���, I <,,-II�.... L--i II I , to terminal 2 I I I I I I I · I . "' control unit 2 r� . . I I · I to terminal M 0-1� : l 1 L Figure 1.5 --- � trol unit M 1-.. : I information __ ______ ___ �n�'::.h�� _ --.: Simple mod el of a centralised switching machine. the d.c. resistance across a pair of w ires, the amplitude of a sinusoidal voltage etc. In message and data systems, special bit patterns are used as signals. Specific techn i ques are discussed later. Whatever t he techn i ques used, the basi c signal types remain the same. Figure 1.6 (called a signal exchange diagram ) shows a basi c set of signals for a terminal that can be used in either a calling or a called mode. In the calling mode the first a ction that the user of the terminal must take is to transmit a seize signal to t he system to indicate that the terminal wishes to make a connection or pass a ( message. The syste m generally responds to the seize by an accept signal for ) reasons which will be come apparent later . The terminal then transm its routing signals and the system responds w ith a variety of status signals, for example: - line busy - line/ree - number invalid - line answered Once the need for the conne ct ion is over, the terminal sends a clear forward s ignal. 'Forward' and 'back' here refer to the direction of traffi c which is deemed 14 INTRODUCTION TO SWITCHING SYSTEMS seize seize eccept eccept routing termine. status (ceiling, Figure 1.6 switching terminel system (celled, clear forward clear forward dear backward clear beckward Ba sic signal exchange diagram. to flow from the originating terminal to the terminating terminal. In the called mode the terminal is sent a seize signal from the system and responds with an accept signal. At the end of the period of communication the t�rm!m�! m�y �e sent a clear forward si�al to indicate the end of the connec tion. In some systems a clear back signal is sent from the called terminal to the system and possibly also from the system to the calling terminal if the user of the calle d terminal is the first to indicate that the communication is finished. For example, the signal sent from a telephone terminal to the local centre when the handset is replaced at the end of a call is regarded as a clear forward signal if the user initiated the call, but a clear back signal if the user was the recipient of the call, even though the signals may take the same electrical form in practice. State transition diagrams. The signal diagram gives the 'alphabet' of the valid signals between two devices. However, it does not indicate what sequences (or 'sentences') are possible or what they mean. The valid sequences and their meaning can be expressed conveniently in what is called a state transition diagram (s.t.d.). This is .such a useful specification and design tool that a set of international standards have been produced [6] for use in telecommunications systems. The concepts underlying an s.t.d. can be explained by reference to the control units in Figure 1.5. These control units may be thought of as existing in a number of stable states such as: - idle - waiting for routing information waiting for answer B A SIC SWITCHING CENTRE MODEL 15 connector (from other diagram) state description state no. state box { event name event box action boxes normal action send signal decision box connectors (to other diagrams) (a) invalid (b) Figure 1.7 Example of state transition diagram symbols. (a) Basic symbols. (b) Multiple decision box. In general a con trol uni t moves from one s ta te to ano ther only because of the arrival of a signal from a tenninal or another con trol uni t. The arrival of a signal is described as an event. When the uni t does move from one s ta te to another i t may perfonn some action such a s opera ting a swi tch or sending a signal to 16 INTRODUCTION TO SWITCHING SYSTEMS another control unit or terminal. The actions performed between one stable state and another are collectively called a task. In many cases the combination of current state and new event defines the task and new state, but in some cases there is a choice of next states. Such a choice depends upon information external to an individual control unit. The most obvious example of this is the choice of next state after the routing signal has been received. A different state is necessary depending upon whether the called tenninal is busy or free. Thus there are four components of a state transition diagram, the symbols for them are shown in Figure 1.7: (a) State boxes. These are labelled with a number and descriptive title. In some cases it is useful to place a pictorial representation of the state in box. (b) Event boxes. The possible (or permitted) events are each drawn as arrow indented boxes, the paths to which come from the state box in question. (c) Action b oxes. Actions are shown by rectangular boxes except for the action of sending a signal to another control unit or tenninal which is (d ) given the special symbol of an arrowed box. Decision boxes. For binary decisions the symbol is the diamond-shaped box as normally used in computer flowcharting. For multiple decisions the extended version shown is used. S.t.d. for the calling states of a simple control system. Figure 1.8 shows a simplified s.t.d. for the states of a control unit involved on an originating call. In the idle state the only valid event is seize. This is acknowledged by an accept signal and the control system moves into state 1, waiting for routing. The event of reception of routing leads to a test of the state of the required terminal. ( The mechanisms for performing this test are discussed later.) If the called terminal is free, the calling terminal is informed, a path is set-up, and the unit moves into state 2, waiting for answer The accept signal from the called terminal causes an answer signal to be sent to the calling terminal, and the control unit moves to stat