UXD5: A Small Digital-Switching Telephone Exchange for Rural Communities

 Part 1-General Description 

J. R. w. AMES , B.SC., M.SC., C'.EN(i., M.l.l:..E., M. ], ELSDEN, B.SC., M. w. HILL, M.SC., C.ENG., M.1.E.E., and P. A. TR U DG E'fT, B.Sc., M.sc., PH.o., c.ENG., M.I.E.E. t 

UDC 621.395.20 621.374  POEEJ, Vol. 73, Jan. 1981 

Picture from   Maintenance News

INTRODUCTION 

Rural areas have always presented particular problems for telephone operating administrations. Scattered communities with low telephone penetration, often coupled with adverse terrain, lead to difficulty in offering an econon1ic and reliable service. In the early days of the UK telephone network, service was provided by small manual boards, usually sited at the local Post Office. During the 1930s it became apparent that these manual boards were becoming increasingly uneconomic, therefore the British Post Office (BPO) introduced a family of small Strowger exchanges to cater for the needs of small rural communities. these exchanges, designated Unit Auto1natic Exchanges (UAXs), were constructed from a series of standard cabinets housed in standard buildings. The standard design of these exchanges led to low installation costs and high reliability, and these exchanges offered to the rural subscriber a service con1parable with that available in towns and cities. UAXs still forn1 a large part of the BPO network in rural areas, having been modified for trunk dialling from 1969. There are over 2000 exchanges in the UK network with Jess than 400 lines, a high proportion of which are of the UAX type. Scotland has a particularly large nun1ber of small exchanges, over 600 of which have less than 200 lines. Many of these small exchanges arc now over 40 years old and are reaching the end of their service lives. Modernization of the BPO network is based on the System X family of exchanges: it is the intention of the BPO to replace all Strowger exchanges by the turn of the century. Initial devclopn1ent of the modernisation programme has necessarily been concentrated on the larger exchanges and, therefore. an intcrin1 progran1n1e is required for the replacen1cnt of UAXs; in particular. the replacement of UAX12s. 'f�c cost of developing a new sn1all exchange would be very high, and low equip1ncnt quantities would lead to high production costs.. These costs have, however, been reduced substantially by adopting a new exchange design based on an existing system Which is scheduled for bulk production; that is, the Monarch 120 PABX1, which can be adapted with n1inin1urn design change to fulfil the operating requiren1ents of a UAXl2. In early-1978, a feasibility study indicated that modification of the Monarch 120 system to provide the facilities of a UAX 12 would be possible, and recommended that a model be built for trial in the public telephone network. A feasibility trial model entered service at Glenkindie, near Aberdeen, on the 31 July 1979. The trial has been successful. and production of the system, known as Unit Exchange Digital No. 5 (UXD5), is scheduled to commence in 1981. 

BACKGROUND TO THE DEVELOPMENT OF THE UXD5

The UAX12 is the smallest of the unit Strowger exchanges still in use; there are over 200 in service, mostly situated in Scotland. Initially, the UAX12 was designed to cater for 100 terminations (subscribers plus junctions), but later development permitted extension to 160 subscribers. Switching is centred on pre-2000 selectors, which act as group selectors for junction calls and as final selectors for terminating calls. The basic UAX12 exchange is housed in three types of standard unit, designated A, B and C. The A and B units contain line circuits, subscribers' meters, line finders and selectors; the C unit contains the main distribution frame (MDF) and common equipment. Other units, housing junction and coin-and-fee checking relay-sets, have been added to provide trunk dialling facilities. Junctions carry mixed level I, 9 and 0 traffic; a discriminating digit is used to segregate ordinary and coin-collecting box (CCB) level 0 traffic at the group switching centre (GSC). A typical exchange of 90 lines comprises a suite of racks approximately 4 111 long and 1 ·8 n1 high. Investigation into possible rr1ethods of replacing the ageing UAX 12 installations included the possibility of using concentrators or network amalgamation. However, it became apparent that the cost involved in changing the established network in ren1otc areas would be substantial. Therefore, the BPO decided to develop a replacement systen1 based on the design of the Monarch 120 PABX. The Monarch 120 systen1 is the new standard rental-range PABX which has been developed by the BPO in conjunction with UK telecommunications manufacturers. The system caters for a maxin1un1 of 120 dial or keyphone extensions, which is approxi1natcly the correct size for a UAX 12 rcplacen1ent. The Monarch 120 system offers many of the facilities needed on a public exchange and its microprocessor controi2 has an inherent flexibility, a feature that has eased the modification process. The UXD5 feasibility trial n1odel was developed to demonstrate that a s1nall public exchange, based on the Monarch 120 design, could be built with minimum change to the basic design. 

The design of the UXD5 was concentrated on meeting 4 main objectives: 

(a) that the exchange would be capable of operating in the same network configuration as a UAX12;

 (b) that, for ease of installation, the system would be of unit design; 

(c) that, for ease of maintenance, the equipment would be provided with advanced diagnostic facilities; and 

(d) that the system design would be flexible to enable the provision of advanced facilities, should they be required. 

ADAPTATION OF THE MONARCH 120 PABX AS A PUBLIC EXCHANGE 

Originally conceived as a PABX, the Monarch 120 lacks certain features normally required for the public network. The main areas that required modification were

1. (a) availability performance and power supply arrangements. 

2. (b) caII-charging facilities,

3. (c) junction signalling facilities, and 

4. (d) operation of CCB services 

Availability Performance and Power Supply 

The reliability of the Monarch 120 system depends on the correct operation of the control shelf and the mains power shelf. A failure in either of these subsystems could disable the whole, or a substantial part, of the exchange. Theoretical calculations have indicated that the mean-time-between failures (MTBF) of the exchange due to either of these causes is approximately 3 · 5 years, excluding failure of the mains supply. While this value of MTBF is appropriate for a sn1all PABX in an office environment, it is not suitable for a ren1ote unattended public exchange; a MTBF in excess of 50 years is required in this application. For the UXD5 exchange, an increase of MTBF to approximately 100 years has been obtained by duplication of the control shelves and by introducing DC-DC converters operating from a standard -50 V supply, which removes direct reliance on the public n1ains supply. One control shelf acts in a standby mode and is activated on detection of a fault in the working control shelf, thereby maintaining service until the faulty control shelf is repaired. Use of the DC-DC converters a!IO\VS a small area of the exchange to be powered from each converter so that, in the event of a converter failure, service to only a limited number of subscribers is affected. 

Call Charging 

The Monarch 120 system offers a selective call-logging facility which does not meet the requirements for public network call-timing and charging. At the time of the UXD5 development the billing arrangen1ents for System X had not been finalized, so it was therefore decided to develop the hardware and software required to operate standard subscribers meters in the usual manner. Charging operates in exactly the same way as a UAX exchange by using local-call timing (LCT) and metering-over-junction (MOJ) pulses from the parent GSC. However, the UXD5 systetn has been designed so that a billing system providing System X data can be easily attached at a later date. 

Junction Signalling 

There is no provision in the Monarch 120 system for public network signals such as MOJ, trunk offer (TKO), and manual hold because the exchange interfaces only with public exchange lines. Hardware and software have therefore been developed to interwork with existing loop disconnect (LD), SSDC2 and SSAC8 signalling systems. A circuit is being developed which, with the aid of a single-chip microprocessor, will be able to work to a wide range of DC signalling systems, including those mentioned above. Description of the UXD5 software and system operation will be the subject of Part 2 of this article, which will be published in the April 1981 issue of the Journal. 

Operation of Coin-Collecting-Box Services No provision was needed in the Monarch 120 system for the control of CCBs. A self-contained CCB for use in the low revenue areas serviced by UXD5 exchanges is not likely to be available for some time, so a special line unit has been developed to handle coin-pulse signals and to control the coin slots of existing pay-on-answer CCBs. Use of this line unit, plus additions to the exchange software, enables UXD5 exchanges to perform the standard coin-and-fee-checking process without the addition of any further hardware. 

Line Shelves

 The line she1ves and backplanes of the UXD5 are identical to those used in the Monarch 120 system. The line cards interface with customers' 2-wire lines and provide digital speech and signalling, at 72 kbit/s, to and from the shelf 1nultiplexers; the signalling and speech are separated in the shelf multiplexers. A line card contains 4 subscribers' line circuits. The MDF incorporates gas-discharge protection devices on each line to prevent lightning surge voltages being passed to the digital circuits. The line shelves house the line units designed for use with CCB lines and junction signalling cards; receivers for multifrequency signailing (SSMF4) telephones can also be accon1- rnodated on the line shelves. Monarch 120 multiplexers are used to multiplex the digitally encoded speech from the 32 ports on a line shelf onto a single 2·048 Mbit/s highway and to perform demultiplexing in the opposite direction of trans1nission. The digitally t:ncodt:d signalling to and froni the 32 ports on a line shelf is carried on a pair of 256 kbit/s highways, one for each direction of transmission. The line shelves also accon1tnodate a test line unit, which provides facilities for the automatic routine testing of the digital circuitry associated with each line and junction card, as well as n1uch of the control shelves; the test line unit design is the san1e as that used for the Monarch 120 system. 

Control Shelves

 Although the control shelves use a high proportion of Monarch 120 circuit cards, the backplane wiring has been modified to allow duplication of the control shelves to meet the availability requirements. Only one of the control shelves is active at a time; the other control shelf is operated in an idle mode, in which it executes a program that checks for faults in its own equipment. The idle shelf takes over the control of the exchange automatically should a fault be detected in the active unit. Each control shelf includes a change over card specially designed for the UXD5. These cards incorporate tri-state buffers which pass clock waveforms, signalling and digitally encoded speech fron1 the operational control shelf to the ribbon cables which interconnect the line and control shelves. The change-over cards on the control shelves are interconnected in such a fashion that only one shelf can have access to the line shelves at any one time. If a fault is detected on the working control shelf, an output from the processor circuitry, known as the watchdog1, causes the change-over card to initiate a change-over to the idle control shelf, but with loss of calls in progress. A routine change-over between control shelves is also made approximately every 24 h, but at a time when no calls are in progress. A data link is provided bet\veen the two control shelves so that changes (such as class-of-service state) made to the database in one shelf can be passed automatically to the other shelf. Digitally encoded speech from the shelf multiplexers is applied to non-blocking tin1e-switches designed for the Monarch 120 system. The time-S\vitch inputs include signals fron1 a digital tones card and a conference unit (currently used in the UXD5 for various standard functions such as tone test and pay tone), as well as signals from the 6 line shelves. Similarly, signalling from the shelf multiplexers is applied to a signalling input card on each control shelf, which is scanned by the processors for signalling information. A signalling output card interfaces between signalling from the processors and the line shelves. The provision of public network signalling and metering has placed an additional scanning load on the central processor unit (CPU); to assist with this load, a pre-processor unit (PPU) has been introduced on each control shelf. Both processors use the same design of board, the differences being implemented by means of links. In each case, an 8085 microprocessor provides the processing power and is provided with an attendant watchdog circuit, random access memory (RAM), read-only memory (ROM) and serial data-handling devices (universal synchronous/asynchronous receiver/transmitter (USART)). Alphanumeric displays on the front of the board are used to provide limited diagnostic information. Primarily, the PPU handles junction scanning and dialled digil n;:cepliun, for which a scan every 8 ms is required; the CPU deals with a slower (128 ms) scan used for the detection of subscribers' loop conditions and for the handling of call processing routines. The use of two processors \\•as necessary to handle the volume of processing work and to provide sufficient serial USART links for purposes such as metering, without requiring major redesign of the Monarch 120 processor board. The two processors on each control shelf are arranged to use co1nmon data and address busses. The PPU has priority and can interrupt the CPU by use of a hold signal. The main program is held on two ROM cards, each capable of storing 48 kbytes2 of inforn1ation; 16 kbytcs of RAM is also provided. Both types of storage use the same cards as those used in the Monarch 120 system. 

Metering 

The meter-driver cards are driven from the PPU over duplicated serial data links controlled by USARTs and are accessible by either control shelf.

 Each meter-driver card controls the operation of up to 60 n1elers. The card carries a single-chip microcomputer (Intel 8748), which receives a request to operate a meter, checks the n1essage, holds the meter operated for the appropriate length of tin1e and checks that the meter is being correctly driven. Relay driver integrated circuits are used to interface between the low voltage microcomputer and the meters, which operate fron1 the - 50 V battery. Up to 4 meter-driver cards can be connected to a bus, which carries n1essages to and fro1n the PPU. When a meter is to be operated, an 8 bit serial n1essage is sent to the bus by the PPU, the first 2 bits of which identify the driver card. The rnicroco1nputers on each of the 4 cards receive the n1essage, but action is taken only on the card addressed by the first 2 bits. The remaining 6 bits of the message identify the meter that is to be operated. The n1icroprocessor decodes the message and first checks the electrical conditions at the meter terminals lo confirm that the meter is not already operated. If this test is successful, a signal is sent to the appropriate meter driver to operate the n1eter and the conditions are again checked, this time to ensure that the 1neter has been driven correctly. After a period of 150 ms, the meter is released and a final check is performed to ensure that the meter is no longer operated. Finally, if all of the checks are successful, the microprocessor returns a copy of the original message to the exchange PPU via the bus to verify that a meter pulse has been applied. The entire meter-driver structure is duplicated (as shown in Fig. I). In the event of a fault on an operational meterdriver card (detected by a wrong or missing reply message), no further messages are sent to that card. Further requests to operate one of the 60 meters thereby affected are routed by the PPU via the second bus to the duplicate standby card, and an alarm is generated. In the unlikely event of a failure in the standby card, no further attempt is made to charge any of the 60 affected customers until maintenance action is taken. At intervals of 24 h, the functions of worker and standby meter-driver cards are interchanged to ensure that hidden faults have not developed in the standby unit. 

PHYSICAL REALIZATION OF THE UXD5 The 150-line exchange is contained in two cabinets, secured together to give an overall size of 1114 mm x 595 mm x 2130 mm. The cabinets arc designated A and B respectively, and contain standard-size shelves which accommodate equipment cards of 318 mm x 203 mm. The equipment cards arc interconnected by means of a backplane in which wiring is wrapped automatically during production. A front view of the cabinets (with covers removed) is given in Fig. 3. 

B-Cabinet Equipment The 8-cabinct (shown on the right-hand side of Fig. 3) houses equipment required to connect the UXD5 to the existing network; that is. the equipment that enables the UXD5 to operate as a direct functional replacement for a UAX 12. The cabinet houses subscribers' meters and their associated duplicated driver cards, an alarm panel, the MDF, miscellaneous relay-sets and the 50 V power distribution fuses. Three pairs of meter-driver cards arc contained in a standard shelf; connexion is made to the meter panels below by multi-way ribbon cable terminated with plugs and sockets. The meters are arranged in 3 panels, each holding 60 meters in four rows of 15. Below the meters is the alarm panel and associated indicator lan1ps. \Nhich report the status of the exchange. Also provided on the panel are buttons to reset the nieter-driver cards after niaintenance work; S\vitches are provided which disable the alarn1 indicators while the la1nps are tested. The conipact MDF is built fron1 the recently introduced Jacks, Test No. 39 and No. 40. Each block carries 100 pairs; those on the line side of the exchange are protected by gasdischarge tubes. An intermediate distribution fran1c is not required since directory-number to equipment-number translation is pcrforn1ed by the exchange software and the exchange is not sensitive to the distribution of traffic load between the line units or shelves. Therefore, line pairs are jumpered fron1 the line side of the MDF directly to the exchange side, where terminations are arranged in equipn1ent nun1ber order. Direct test access to each line is available at the MDF as each pair can be split by a specially designed plug inserted in the tern1inal block. Power fro111 the exchange battery enters the UXD5 at the botton1 of the 8-cabinet and is distributed as required via fuses. Indications of fuse failure are connected through the processor in the A-cabinet and an alarn1 is raised when a fuse fails. Finally, space is provided for up to 3 jack-in relay-sets. A WB700 receiver is provided to receive local call-ti1ning pulses fron1 the controlling GSC. To allow full flexibility of use, all connexions to the relay-set sockets appear on a 111iscellaneous connexion strip 1nounted in the rear of the cabinet. 

FACILITIES

 The UXD5  provides normal Strowger facilities, with UAX12 variations such as co111bined level 1, 9 and 0 junction working, tron1bone junction testing and ren1ote test nun1ber. Loop disconnect, SSDC2 and SSAC8 facilities are available for junction signalling; the resistance limit for incon1ing loop-disconnect signalling is 2000 �-l. All level I, 9 and 0 traffic is routed to the parent GSC, and side routcings are available on level 8. The nu1nbcring range used on trial n1odels has generally conformed with the UAX 12 nun1bering sche1ne; however, UXDS is capable of operating within a Iinked-nun1bering scheme. The traffic capacity of the exchange is limited by the nun1ber of call records to 23 erlangs while providing a grade of-service of 0·01. The occupancy of the control sets a lin1it of 1700 busy-hour call atle111pts. However. the ti111e-switch is non-blocking. and has no effect on traffic capacity. The trans1nission aspects of the system have been designed, as far as possible, to keep speech levels at the GSC si1nilar to those fro111 a UAX12. There are, however, two restraints upon transn1ission through the exchange. Firstly, the inco1ning analogue signal niust be adjusted to ensure that the signal presented to the CODEC docs not cause overloading; secondly, since UXD5 is 4-v.·ire sv..1itched, the 4-wirc loop n1ust be unconditionally stable .. ·rhc exchange uses a 25 mA constant-current feed on subscribers" lines and, as the telephone regulator is inoperative at this level of current, a gain adjustn1ent is provided on the line card to cater for short lines of up to 4 dB line Joss and long lines up to 10 dB loss. Service can be provided to long lines exceeding 10 dB loss by using a modified line-card. At present, the DC signalling lin1it for subscribers' lines is 1500 fl, excluding the telephone instru111ent. Subscribers' class-of-service status is stored in the exchange configuration database, which can be read and 111odified using a teletype. Access to the database is restricted by n1eans of passwords. New lines can be provided sin1ply by plugging-in extra line cards, providing MDF jun1pering and activating the lines in the database. Maintenance and diagnostic facilities can be accessed using the san1e teletype, and will generally isolate faults down to card level however. as these automatic tests are not absolutely certain of isolating all faults, confirn1- ation by maintenance staff is required. Results from the tests arc stored in a fault record, which can be read as required by maintenance staff. The diagnostic facilities also allow the n1aintenance staff to run particular tests on suspect equipn1cnt. Initially, production UXD5 exchanges will be provided with adaptors for 16 shared-service lines, but the space occupied by the adaptors v.·ill possibly be needed for new equipn1cnt providing advanced facilities as shared service is phased out. 

FEASIBILITY STUDY TRIAL

 The UAX12 site chosen for the feasibility study is at Glenkindie, in the Aberdeen area. The exchange serves a rural community and has 93 exclusive lines, 3 CCBs. and 8 bothway junctions to the GSC at Alford. The connexions include both very short lines and some lines which are beyond the present transn1ission limits, but which arc within proposed new lin1its. 6·85m For the purposes of the feasibility study, the existing UAX 12 is retained on site. although some rearrangen1ent of the UAX I 2 racks was necessary to acco1nn1odatc both exchanges in the existing building (see Fig. 4). The ability to locate both exchanges in the one room indicates just how small the UXDS is in relation to its predecessor. A multiple relay switch is provided so that either the UXDS or the UAXl2 can be connected to provide service. The UAX 12 is returned to service automatically should the UXD5 be rendered inoperable by failure of both control shelves or by failure of the power to one of the DC- DC converters supplying the line shelves. For the feasibility study trial, it was necessary to increase, the capacity of the exchange batteries to supply the joint power demand when the UXD5 carries no traffic and the UAXl2 carries all traffic. If the mains supply fails for longer than 5 1nin, the UXD.5 is switched off to conserve battery power. ln this case, the UAX12 provides service. Urgent alarms in the UXD5, or a change-over to the. UAX12, are signaled remotely by a change to the number unobtainable (NU) tone on the alarm-test number. 

The trial was started on 31 July 1979. Some initial deficiencies were discovered during the early stages of the trial, but these have been easily corrected. In particular, the need to reduce the susceptibility of the line cards to lightning was highlighted, and improved protection was provided and will be provided on all future models. 

FURTHER DEVELOPMENT OF UXD5

 The BPO is now producing a further 12 exchanges for laboratory and network trials. These arc based on the G lenkindie design, but some minor changes have been in1plcmented as a result of experience with lhe trial n1odel. Nine of these exchanges arc being installed in the Dundee and Edinburgh telephone areas for evaluation trials. A contract has been placed with Plessey Telecommunication Ltd (PTL) for industrialization of UXD5 design and the production of a nun1ber of prototype exchanges. It is expected that production exchanges will become available during 1981. A joint BPO/PTL study has been set up to identify possible enhance1nents for both the home and export markets. A substantial export market currently exists for a very small exchange of this type, and the initial phase of the study included an asse!'lsment of export requirements. Further development of UXD5 is envisaged in the following areas: (a) To reap maximum benefit fron1 export opportunities and for additional flexibility within lhe home market, it is proposed to increase the maximum size of UXD5 to 600 lines. (b) Consideration is being given to the development of additional services; nan1ely, abbreviated dialling, automatic advice of can duration and charge, automatic alarm-call, call barring, caII diversion, call-waiting indication, repeat last call and 3-party calling. The Monarch 120 system offers a range of supplementary services \\'hich \vill form the basis for equivalent facilities on UXD5. (c) Advanced signalling development is in hand to provide CClTT R2 signalling and direct connexion to 30-channel pulse-code modulation systems. In addition, the development of co1nn1on-channel signalling is under consideration. (d) The design is being enhanced to provide advanced service facilities, including remote print-outs and remote interrogation of niaintenance and diagnostics. Advanced traffic-recording facilities are being considered for development at a later stage. (e) An itcn1izcd billing system is under development, and ren1ote read-out facilities for both bulk and itemized billing systems will be provided. 

SUMMARY 

The UXD5 is a small, high-technology telephone exchange designed for use in rural areas. It is of extremely economic design due to its comn1onality with the Monarch 120 PABX; approximately 75%  of the cards in a fully equipped UXD5 are the same as those used in the Monarch 120 system. The design is highly flexible as the exchange is software controlled, and a high availability is offered due to duplication of the control and switching equipment. The unit design of UXD5 leads to ease of installation, and sophisticated diagnostic facilities are provided to assist maintenance. 

A feasibility trial of UXD5 has indicated that the syste1n has considerable potential and an additional 9 models are now being installed in the BPO network for further evaluation of the design. 

The design of UXD5 is now being enhanced for export purposes and to provide advanced facilities for the home market. Production exchanges are scheduled to be available during 1981. 

References 1 POTTER, A. R. Monarch 120-A New Digital PABX. POEEJ, Vol. 73, p. 14, Apr. 1980. 2 ToNGE, J. D., GAUNT, D. L., and KE:-.!DALL, J. P. Programmable Logic and Microprocessors. POEEJ, Vol. 70, p. 136, Oct. 1977. 3 WALTERS, R. E., PARK, 1. D. C., and COLLINS, P. E. Microprocessor Software Development for a Small Telephone Exchange. Microsojtware, SERT Symposium 1980. 

UXD5: A Small Digital-Switching Telephone Exchange for Rural Communities

Part 2-System Software and Operation

J.R. W. AMES, B.SC., M.sc., c.ENG., M.l.E.E., M. J. ELSDEN, B.SC., M. W. HILL, M.SC., C.ENG., M.I.E. E.,

and P. A. TRUDGETT, B.SC., M.SC., PH.D., C.ENG., M.I.E. E. T

Messrs Ames, Hill and Trudgett are with the Research Department, Telecommunications Headquarters. Mr Elsden Is With he

Exchange System Department, Telecommunications Headquarters

Ames, . R. W., et al. 

UDC 621.395.2: 621.374

The UXD5 is a digital-switching telephone exchange designed for use in rural areas. The exchange is intended as a replacement for Strowger Unit Automatic Exchanges at present installed in the UK telephone network.

Part 1 of this article*gave a general description of the UXDS design and its application; Part 2 describes the system software and operational aspects.

Introduction

Part 1 of this article* gave a general description of the Unit Exchange Digital No. 5 (UXDS), which is a digital-switching telephone exchange for use in rural areas where telephone penetration is small. The UXD5 design is based on the Monarch 120 PABX!, and the British Post Office (BPO) intend to use the UXD5 as a replacement for the Strowger Unit Automatic Exchanges (UAXs) now in service in the UK telephone network; in particular, for the replacement of UAX12.

Description of the system is completed in this article, which covers the system software, the interaction of software and hardware and the system operation. The system operation is described by examining the progress of a local-call connection  through the exchange.

Operating System

The operating system is a program that runs continuously, except when stimulated to call and execute other programs.

The operating system exercises overall control of these other programs. The operating system is stimulated by either of the following signal states:

(a) Pulses known as interrupts are sent to the operating system at regular time-intervals. In the UXDS, these interrupts occur at 100 ms and 8 ms intervals. The interrupts force the microprocessor to stop its current task and execute a fixed routine in the operating system. This routine contains calls to programs known as foreground processes.

(b) Messages are passed to the operating system from programs that cause the operating system to run other

programs. These messages comprise a common block of memory, 8 bytes long, shared by the operating system and the

sending program (see Fig. 6). To send a message, the block is filled with the necessary data, and a call is made to the parti-

cular entry point in the operating system that deals with messages. All programs started in this way are known as background processes. A background process can be started by a message from

(i) another background process,

(ii) itself,

(ii) a foreground process, or

(iv) the operating system (only when the exchange is first activated).

The operating system performs several other functions to help in the control of the processes:

(a) Using the 100 ms interrupt signal, the operating system acts as a real-time clock and keeps a count of the date and

time. Any process can obtain the time by sending a special message, which is returned by the operating system with the

date and time appended.

(b) A time-delay function is incorporated so that any process which passes a message to the operating system can

be restarted after a fixed-time delay. The delay period is set by the calling process.

c)Messages to each process are queued so that no messages are lost during periods of high exchange activity.

d)The operating system allocates priorities to each background process so that those processes that have a more urgent function in the exchange are always run when required, to the exclusion of the other processes. 

(e)The faster interrupts are controlled in such a manner that they can always interrupt the slower interrupts, but not vice-versa.

The operating system examines all messages and checks that they have the correct format; if a format is incorrect, the message is stopped and stored in a fault record.

Control Processes

The control processes are divided into several groups according to function. Any one group can have a maximum of one background process, but may have several foreground processes running at different intervals. Processes in the same group share common data-areas, so that data can be saved and passed between processes without involving the operating system.

Exchange Scan (XSCAN) 

The exchange-scan (XSCAN) process group has elements in both foreground and background program areas. The function of the XSCAN process is to interface between (a) inputs from the signalling-input card, outputs to signalling-output card and the time-switch, and 

(b) those processes which perform the telephony functions.

The foreground XSCAN processes are concerned mainly with extracting relevant data from the signalling-input card and the sending of messages to the call-processing functions when a significant change has occurred.

The XSCAN process has two foreground processes:

(a) the slow process (scan 1), which runs every 128 ms (16 x 8 ms interrupts) and examines, via the signalling-input card, all ports on the exchange each time it runs, and 

(b) the 8 ms process (scan 2), which is split into several parts.

The scan 2 sub-processes run at multiples of the 8 ms interrupt (1, 2 and 4) and scan selectively those ports which are liable to have fast condition changes: for example, a subscriber's line during dialling, a coin-collecting box line during the progress of a call when coin pulses can arrive at any time, and an incoming junction circuit when idle since, at any time, pulses must be detected without delay. The XSCAN background process does not scan; its main purpose is to handle the output functions from the time-switch or the signalling-output card.

The XSCAN processes are linked by a storage area in which 5 bytes of random-access memory (RAM) are reserved for each port on the system; these 5 bytes are known as the handler record. The first byte of cach record contains the state of the relevant port, so that the foreground XSCAN processes know which changes from the signaling-input card are relevant. The state of the port can be changed by other process groups by sending a message to the XSCAN background process.

Call Processing

The call-processing group has only one background process,

The call-processing group has only one background process, known as call processing (CPRO). The CPRO acts on messages from the XSCAN process and controls each individual exchange call.  Associated with the CPRO are data arrays called call records

Charge Process

The charge-process group has members in 3 sections: 

background, 128 ms scan (Charge 1 in scan 1) 

and 8 ms scan (Charge 2 in scan 2). 

The group controls all metering functions in the exchange when instructed via messages from the CPRO or the XSCAN process.

The 8 ms scan 2 process controls asynchronous serial communication channels which instruct the meter control to step a meter. The process also verifies that this has been done by receiving back the identical message. Also, by means of inputs from the exchange local-call timing equipment, it decodes local-call ordinary-subscriber charge timing and local-call coin-collecting box charge timing. The scan 1 process maintains lists of the calls in progress and directs scan 2 to send the appropriate meter-control signals over the serial channel. The lists are updated and checked by the background process in response to messages from the CPRO.

The background charge process is instructed when a call has been established and when it is completed. Acting on this information, the background charge process uses local-call timing, unless a message comes from the XSCAN process indicating that metering-over-junction (MOJ) is valid in this case.

Maintenance and Diagnostics

The maintenance-and-diagnostics process has members in

3 sections: 

• background, 

• 128 ms (scan 1 )

• and 8 ms (scan 2)

The process works in two modes:

(a) it scans all cards in the exchange regularly to verify, as far as is possible, that they are functional, and

(b) it checks individual cards, under the control of the maintenance staff.

Under mode (a), the maintenance-and-diagnostics process runs at 14 s intervals, carrying out a different test each time.

Thus, the whole exchange is tested every few hours, and the common-function cards are tested at regular intervals within this period. 

Under mode (b), messages are sent from the CPRO to start a particular test.

The maintenance-and-diagnostics background process is given a lower priority than either the CPRO or the charge background process, by this means, in a fully-loaded and busy exchange, the operating system can stop the schedule of maintenance and diagnostic functions and can thus prevent unnecessary congestion in the system.

List

The list process group has processes in the background.

128 ms ( List 1) scan and 8 ms scan sections ( List 2) . 

It provides an interface to the maintenance teletype via a standard CCITT V24 serial link. By means of inter-process messages, the list process is instructed to print-out relevant data, it can also pass input data from the maintenance teletype to the appropriate process. 

The list process has the lowest priority of any of the groups of processes.

Exchange Configuration Data Base

All fixed information about individual ports on the exchange is held in the exchange configuration data base, and the information can be accessed by any process. The data base is copied from a removable read-only memory ((ROM) to a battery-backed RAM on initial activation of the system. The data in the RAM data base can be changed selectively from the maintenance teletype only after the relevant security procedures have been followed. At any time, on instruction from the maintenance teletype, the RAM data base can be copied to a new ROM.

The types of data held in the data base include: 

• directory-number (DN) to equipment-number (EN) translations: 

• EN to port-type tables; 

• class-of-service (COS) tables; 

• dialled number to call-type table (for example, an STD call if the first digit dialled from a subscriber is 0); 

• lists of incoming and outgoing junctions, 

• and PABX lines (for hunting for a free port). 

There are several more tables of data of a more detailed nature.

Interworking of the Central Processor Unit and the Pre-Processor Unit

The software is divided between two processors: the central processor unit (CPU) and the pre-processor unit (PPU). The PPU operates the exchange scan 2, 8 ms sub-process and the charge scan 2 process. The code for these processes is on the PPU board; accesses to the main backplane data and the address buses are necessary only from the PPU when input, output or common data areas need to be accessed.

The common data-area is contiguous, and the data types are split into two forms: those that are written to by either the CPU or the PPU, and those that are written to by both processors. Data that can be written to by both processors (for example, the handler records, as used by the XSCAN process) needs to be protected against errors caused by read-modify- write commands being attempted by both processors simultaneously. This protection is provided by using two fiags held as bytes in the common data area, so that a processor can check the other processor's flag before writing to its own flag to signify that it is accessing the data. A final check is made on the other processor's flag before proceeding. Apart from this protection sequence, program modules can be moved freely between the PPU and the CPU. Owing to the limited number of accesses made to the main bus by the PPU, the CPU is slowed down by less than 5

SYSTEM OPERATION

To demonstrate the operation of the system, the progress of a local-call connexion through the exchange is described; the description indicates how the hardware and software components of the exchange interact during the progress of the call.

Software Overview

Two programming languages are used to write the software. For the more time-critical and input/output processes. assembler language is used. This is a set of statements that is related directly to the final machine code required; therefore, the processor. The foreground processes and the operating system are written in assembler language. Less time-critical processes are written in a high-level language called CORAL, which uses English-type statements and is therefore easier to understand and write. The programmer does not need program to know the machine code that this will language produce. The background processes are written in CORAL.

The total software requires 68 kbytes of program code for the main processor and 6 kbytes of program code for the PPU. The CPU also has 16 kbytes of RAM, of which 1-5 kbytes are used as a common data area between the two processors. There are 2 kbytes of address space reserved for input and output functions.

Own-Exchange Call

The progress of a call through the exchange, from subscriber 212 to subscriber 202, is now considered. Initially, the 212 line is in the idle state and is examined every 128 ms by the XSCAN process. When the subscriber lifts his receiver the change in loop status is detected by the line card, and the appropriate bits in the signalling input are changed by means of the signalling message stream, which flows from the line card via the shelf multiplexer. On the next 128 ms scan, the change in status is noted by the XSCAN process and an inter-process message is generated to inform the CPRO that line 212 is originating a new calI. The messages refer to exchange ports by equipment number (EN) rather than by directory number (DN). The EN represents the physical location of the port, and can have any desired DN associated with it by means of a DN-EN translation table.

The progress of the call is indicated in Table 1, which lists all the inter-process messages generated during the call. On receipt of the message from the CPRO to set the state of line 212 (stored in the handler record) to dialling enabled, the XSCAN process begins to scan the appropriate signaling-in Jocation every 8 ms in order to analyse the dialled digits. The final message in the initial sequence tells the XSCAN process (which controls the digital-switch hardware) to connect line 212 to the digitally-generated dial tone from the tones card. By this time, the CPRO has assigned a small area of RAM storage (the call record) exclusively to this call. Full details of the progress of the call (digits dialled, path set-up etc.) are stored in the call record. The CPRO is therefore able to handle a  large number of different calls in turn, since the state of each call is remembered and can be recalled when action is required.

The first digit (2) is dialled and it is analysed by the 8 ms Scan. When the inter-digit pause is detected, the CPRO is informed that a 2 has been dialled. On receipt of this message the CPRO fetches the call record, stores the 2, notes that it wa the first digit dialled and sends a message to the XSCAN process insiructing it to disconnect dial tone and connect

Silence before storing the modified call record for later use, The second digit is analysed and sent to the CPRO, as is the third digit. On receipt of the third digit the CPRO is able, in this case, to determine the destination of the call. The DN-EN table is accessed to find the Jocation of the appropriale Iine circuit in the exchange (in this case the EN is 69); the Sequence of events is shown in Table I. The CPRO instructs the XSCAN process to set the handler-record state of 212 to awaiting answer and to mark line 202. If line 202 is free, the mark wilI be successful and, on receipt of this message, the CPRO instructs the XSCAN process to connect digitally-generated ringing tone to line 212, while connecting ringing current to line 202. Because the ringing supply is not cadenced by the ringing generator, the ringing supply relay is operated and released under control of the XSCAN process via the signalling-output card to send ringing current immediately on line 202; this is followed by the standard public-exchange cadence.

On receipt of called-subscriber answered condition (CSA), lines 202 and 212 are connected together via the digital switch to allow conversation to take place, and local-call charging begins. When the calling subscriber clears, the on-hook condition is detected by the XSCAN process via the signalling-input card; the call path is then disconnected and charging is stopped.

A similar mechanism is used to control other types of call.

For a call originating from a coin-collecting box, the coin-and-fee checking function is performed by the processor. In this case, the XSCAN process has to control the operation of the coin slots and to send a message to the CPRO when a coin is inserted. The CPRO controls the duration of the call and signals to the XSCAN process to terminate the call when the call time exceeds that allowed for the money that has been inserted. Compensatory time-allowances are made as necessary; for example, for the time taken for a coin to fall past the slot lock. Messages are sent to the charge process to record the value of the coins used.

Incoming and outgoing junction calls need a separate call control mechanism from that used for own-exchange calls.

Metering-over-junction (MOJ), CSA and backward hold are recognized by the XSCAN process from an outgoing junction, and trunk-offer (TKO) and howler controls are recognized from an incoming junction. For outgoing junctions, the XSCAN process also needs to know from the CPRO whether the call is a manual-board call, a level-9 call or an STD call, as the return and forward signalling requirements can be different in each case. The CPRO detects whether the calling party releases the call, or if it is backward held. 

Other types of call are catered for; for example, 999 calls, for all these types of call are catered for by different routines within the CPRO.

CONCLUSION

The UXD5 has been developed in response to an urgent for a modern system to replace ageing rural exchanges, mainly to be found in Scotland. Microprocessor technology has enabled the system to be developed as a small but sophisticated stand-alone exchange. Economies have been achieved by basing the design on the Monarch 1 20 PABX, which has provided both common hardware and a base from which to develop the system software.

The requirement for 2 versions of the UXD5 has been identified; the UXD5A, designed as a direct replacement for the UAX12, is a 150-line system and the UXD5B, being designed for both the BPO network and the export market, will have a maximum of 600 lines together with a range of enhanced facilities. Production versions of the UXD5A

exchanges are scheduled to be available in late 1981; the UXD5B, which will supersede the UXD5A, is expected during 1982.

GEC UXD5