GB2207835A - Message based signalling system architecture - Google Patents

Abstract

A telecommunication system interconnects a number of users such as terminals, exchanges and concentrators via interface units (TAP), which interface units are interconnected by point to point links (T) to produce a network which may be local, national or even international. The users may use different signalling protocols (e.g. DASS-2, DPNS and I.4xx); to avoid having to translate all such messages, if a signalling message passes via the system between two of its interfaces it is provided with a routing header. This header is attached to the message when it enters the system and is removed when it leaves the system. Thus the system is fully transparent to such signalling messages. Within the system, signalling is conveyed as fixed length packets, so that a long message is sent as two or more such packets. <IMAGE>

Description

Message Based Signalling System Architecture This invention relates to a telecommunication system in which a number of types of network user, some using different signalling protocols, are interconnected.

Such a system enables analogue, digital, data, message and ISDN based users of a telecommunications network to access both public and private network switching and handling facilities flexibly. In such a system it is possible for sub-system circuits and facilities to be re-assigned from time to time to accommodate varying patterns of user traffic and the use of private and public network facilities.

An object of the invention is to provide a system which satisfies the above desiderata.

According to the invention there is provided a telecommunication system which interconnects G number of interface units each associated with one or more of a number of users of the system, wherein connections may be set up via the system between the users, wherein signalling relating to the connections to be set up via the system uses a plurality of different signalling protocols, wherein when a message using one of said protocols arrives at a said interface unit that message is provided with a system header appropriate to the passage of that' message through the system, and wherein when a said signalling message with a said system header reaches the interface unit appropriate to its destination the system header is removed from thatmessage and the message is routed from the said interface unit to its destination, so that the system is fully transparent to that message.

An embodiment of the invention will now be described with reference to the accompanying drawings, in which: Figure 1 is a schematic block diagram of a telecommunication system emDodying the invention. This diagram is relevant to the signalling used in the system, and does not include details of the actual connections to which the signalling relates.

Figures 2 to 7 are explanatory diagrams representing frame and bit structures such as are encountered in a system such as that shown in Figure 1.

Figure 8 shows schematically part of a system environment.

Figures 9 and 10 relate to certain signalling message routing problems.

Figure 11 relates to interfacing at the input ena.

Tne system enables analogue, digital, data, message and ISDN (Integrated Services Digital Network) based users of a network to access both public and private network switching and handling facilities flexibly. Under control of a management sub-system, circuits and facilities may be re-assigned from time to time to accommodate varying patterns of user traffic and the use of the network facilities.

The system shown in Figure 1 includes transmission links T, each providing for digital transmission at the systems bit rate, and higher orders. These are mainly optical fibre links, but may include metallic carriers such as coaxial cable, or even radio lInks. The system nodes are interconnected by the links T, and are called TAP's (Transmission Articulation Points), so that the TAP's and the T's together form a national or even international local flexible transmission network.

The transmission network nodes serve a variety of users, as follows: (a) TERM; individual terminals (analogue telephone or data) each connected to a TAP via a link A for analogue transmission by pair or quad with analogue/digital conversion the TAP.

(b) ISDN TERM; ISDN individual terminals each connected to a TAP by a digital link, in this case 144kbit/sec for basic access.

(c) ISPBX; an integrated services digital PBX with connection P for private and public network access.

(d) LE; local digital exchange, part of the PISTE (Public Switched Telephone Network) and connected to a TAP by a T link.

(e) TKE; digital trunk exchange in the POSTS.

(f) CONC; concentrator in the PSTN.

(g) PABX; analogue or digital PABX connected by the appropriate link type to a TAP.

The system has Network Management units each serving a number of TAPfi's and connected thereto by digital network management links NML. There are also higher layer management systems, connected to the network management units by further NML's The system provides for flexible multiplexing at primary and higher rates, circuit tgrooming" to ensure reasonable "fill" of the multiplexes, automated cross-connect facilities for traffic circuits and sub-rate channel-associated signalling (CAS), interfacing to users and public and private network facilities, and the general handling and routing of message based signalling systems between users and their interfaces with the networks.

The block ISDN represents an ISDN associated with the system, PSPDN is a public switched data network also associated with the system, and ETC represents any other network associated with the system.

The different forms of network served mean that regard has to be had to handling different message-based signalling systems used for call setting, control andrelease. Various message based systems using common channel facilities may be encountered. Such systems as in-band single or multiple frequency, and CAS in timeslot 16 of a multiplex are not difficult to deal with. Thus in-band is routed with its data and CAS in TS16 is routed with its traffic channel using a cross-connect switch for the sub-rate CAS channels at 2KDit/sec. CCITT No. 7 signalling is in most cases associated with the same multiplexed groups of traffic circuits as a TS16 channel, so it also should not cause difficulty.

The main problem is the handling and routing of common channel message Dased signalling systems.

Examples include the CCIT' -T.4xx systems of which a number of variants exist, ane the BT DASS-2/DPNSS systems which differ from the I.4xx family. Hence we have to deal with DASS-2, DPNSS, I.4xx and developments thereof.

Thus a form of protocol conversion is usedr in which the various message based protocols at the peripherals are converted to a common base for inter-node (TAP) message routing. Complete conversion is not used because of the variety of such protocols.

Hence a system protocol is provided by adding a special header on entry and removing it on leaving. These are called wrapping and unwrapping respectively. Such a header contains the information needed to route the message via the system allowing the whole of that message in its original protocol to be routed without modification.

Note that if interworking between two different peripheral message-based signalling systems is needed, specific translation is done as a separate operation, possibly as a system peripheral function. One example of this is our own IDA (Integrated Data Access) Mux in which users with single line I4xx are connected to users with primary rate DASS-2.

In the DASS-2 and DPNSS protocols the first 1 or 2 octets after inter-frame flags contain the address, and here I.4xx is similar.

In DASS-2/DPNSS and I.4xx, the address information of the signalling frames is in terms of the endpoint to which the link or channel operates. In the fixed network, this implies â user at one end and a switching point or user at the other, i.e. a single link.

Consider first CCITT I.4xx, see Fig. 2. Here SAPI is one octet which is a Service Access Point Identifier which is the higher order octet, while TEI is an octet forming a Terminal Endpoint Identifier. C is one or two control octets, I is the information field (N octets), FCS is the frame check sequence (two octets) and the flag is a unique octet 01111110. Between any two signalling frames there are two Flag sequences.

When no signalling frames are being sent, there is no transmission. The change from the opening Flag sequence to other sequences indicates the start of a signalling message frame. The HDLC protocol is applied to the signalling frames.

The system routing information is introduced between the end of the Flag sequence and the start of the signalling frame proper, and allows it to be transported across the system. Typically, the routing information occupies two octets, but in some areas only one octet, or more than two octets may be required.

Also it is possible that a minimum of only one, rather than two Flags might be allowed between message frames, minimising overhead.

The originating TAP analyses the original message to determine the TEI value in the second octet of the frame. This is then correlated with the physical incoming link identity information and connectivity intormation in the Management System at the originating TAP, to provide the routing octets. These are inserted in front of the original message and, by the message routing function, enable the wrapped signalling message to be transmitted over the common channel link to the next TAP, which may be an intermediate or exit TAP according to the routing.

At an intermediate TAP, the routing infor:ation octets of a frame are examined and correlated with the physical identity of the incoming link and the connectivity information in the local Management System to a substitute of the original routing octets with new octets to enable the original frame to be routed to a further intermediate TAP (where the re-routing process may continue) or to an exit TA? and thence, with final processing, to the required user, or public or private switch as appropriate.

At the exit TAP it is simple to route to the designated outlet and strip off the routing octets, forwarding the original I.4xx message frames to the appropriate destination based on information derived from the local System Management.

Consider now the routing of DASS-2/DPNSS message frame sequences, Fig. 3. Here A is the Address field (one or two octets), C is the control field (one octet), I is the information field (N octets - not used on response frames), FCS is frame check sequence (two octets) and finally there is the flag. Here there is a minimum of one Flag sequence (01111110) between consecutive frames, although, in the absence of signalling frames, there is an idle pattern of contiguous Flags.

Again, at the point of entry, every message is examined to determine its Address. As before, the Address information plus information from the Management System constructs routing octets. The remainder of the routing to the outlet to the destination is done on the examination of the system routing octets, and the substitution of these at intermediate TAPs as in the I.4xx case and the final operation of stripping off the routing octets and replacing them with Flag sequences on the outgoing links to the peripherals.

Thus the two forms of message based signalling can be preserved by providing a wrapping designating routing within the system. In principle, other message based signalling systems can be accommodated on the same basis, provided initial analysis can be done at the point of entry.

The routing relies upon a Management System sub-system which contains full routing and connectivity tables, the related routing octets for TAP to TAP routing, and the ability to change this information to provide the flexible group and circuit routing required i.e. flexiDle grooming and consolidation.

We now consider the number and structure of the added routing octets, Fig. 4. Simple coding of an octet can give 256 addresses, so a single un-constrained routing octet would enable a TAP to differentiate between 256 signalling relationships. This may be insufficient in view of the large sizes of multiplexed circuit groups envisaged. In addition, constraints have to be applied to the routing octet structure.

The routing octet should not be confused with Flags. Hence the allowable values must be constrained to limit the number of contiguous '1' bits to not more than five, using "0" bit insertion if needed.

Thus the routing octets need to provide coding to distinguish them from the Flag pattern, and large address field for a large number of signalling relationships.

Fig. 4 shows one coding using two octets for routing. In the first octet, bit 6, and in the second octets, bits 4 and 8 are set to Binary "O" to prevent Flag imitation and to ensure that combining the second routing octet with the next octet of the message cannot violate the 5 consecutive Binary "l"s rule by taking the last bit of the second routing octet with the first 5 Binary plus of the next octet. The Z bits may be Binary "1"s and "0"s to indicate the identity of aspecific signalling relationship. The coding scheme shown gives up to 8192 such relationships.

The resulting signalling frame forms that can occur are shown in Fig. 5 for I.xx signalling and Fig.

6 for DASS-2/DPNSS signaling, the routing octets being in each case designated El or E2.

Further, the signalling frames due to the wrapping are transmitted link by link through the network without HDLC protocol operation. Thus, the network appears transparent to the message based signalling systems, since routing is performed entirely on the added routing octets.

Some delay to the original messages occurs due to the original message analysis and the introduction of routing octets and terminating Flag sequence. Thus at the point of entry, the signalling frames need to be buffered to provide the appropriate delay while the routing octets are introduced. It is also possible that with circuit consolidation at the same point, or other intermediate and exit points, some delay in forwarding the signalling frames results.

In spite of the need for message analysis at entry points, this reduces the intermediate and exit point processing to an acceptable level; routing is based on a single process of examining the routing octets, rather than having to route on the basis of analysing the individual, and different message based headers of the peripherals at each TAP through which they pass, together with a possible need to terminate and re-initiate the LAPs (Link Access Protocols) at each TAP.

Finally, in the absence of signalling messages, the common channel signalling facility transmits an idle" pattern of Flag sequences.

Recognition of the flag sequence enables both octet and frame synchronising.

In an improved version, to simplify the handling of signalling frames, a standard frame length is adopted that accommodates the commonest sizes of signalling messages. This is defined as G "Mini-Packet". A terminating Flag per frame is no longer essential, but may be used in single or multiple form purely as a "Mini-Packet Filler".

While DASS-2/DPNSS allows for frames up to 45 octets in length, and I.4xx allows for frames up to 260 octets in length, examination of typicai message frame lengths for both systems suggests that a frame length of about 30 octets caters for most messages.

A mini-packet length of 32 octets, including two routing octets, has been chosen, which satisfies most signalling requirements.

From the exit processing point of view, it is desirable to indicate in the routing octets whether the signalling message is within a single system frame, or is multi-framed. This is not needed where the signalling frame used corresponds with the signalling frame size between the system and its peripherals.

However, it is needed when the mini- frame is shorter than that signalling frame, in which case the original message has to be segmented into two or more mini-packets.

Fig. 7 shows the Mini-Packet frame format for complete and segmented I4xx signalling frames. Here (a) is a complete signalling message in one frame, (b) is a frame with the initial part of a message, (c) is a frame with an intermediate part, and (d) is a frame with a final part. The arrangement for DASS-2/DPNSS is similar.

The mini-packets are transmitted continuously and in a synchronised sequential fashion across the network. Thus, a Mini-Packet is always transmitted across a link, even if there is no signalling, in which case it contains an idle pattern, such as alternating "l"s and 'O"s in the 30 octets normally occupied by signalling. This pattern is not mandatory, and ispurely an example of a "filler" for the message carrying section of the Mini-Packet. The distinction between a Mini-Packet with a valid signalling message or s segment thereof, and one with no signalling information is made in the routing octets of the header. A routing octet witn valid signalling in the Mini-Packet has its first bit of the first octet set 1 if it is the last frame of the message and to O i it is not tne last frame.The other 15 bits are all added for routing information.

Where there is no valid signalling content, the routing octets nave the first octet set to i1111110 and the second octet set to a i010... pattern, the seven "l"s provide a unique frame alignment pattern. Other variants are possible, but it should not be possible for tne routing octets and the idle pattern to be simulated by a message observing EDLC rules regarding no more than six consecutive "1"s, 1.e. in both information coding and Flags taken together. Note that the coding of the routing octets when there is a valid message allows for 32767 signalling relationships on the basis that the coding for the routing octets when there is no valid signalling content is disallowed.Note also that in this latter case, the 010101010 pattern provides a signal rich in clock information due to its alternating "O"/"i" characteristic, which enchances the synchronisation capability of the overall system.

The operation of the system of mini-packet message based signalling will now be described.

Fig. 8 shows a typical arrangement of a network with 4 TAPs. For the example described, TAPs 1 and 4 handle the signalling input and output respectively to and from the peripherals, and TAPs 2 and 3 provide intermediate routing for both traffic circuits and signalling.The connections IP1 and IP4 to TAP1 and TAP4 convey traffic circuits and common channel signalling to and from peripherals. Although we have described uni-directional signal flow, this is for ease of description as all signalling and traffic carrying circuits are bi-directional.

The inter-TAP links carry one or more multiplexed groups of digital circuits, each with its own common channel signalling (CCS), as represented by the dotted lines labelled (1), (2) and (3). Thus for each multiplexed group there is a signalling relationship between the main traffic carrying circuits and the CCS.

At any of the TAPs circuit tgroominat and consolidation can occur, as can routing. Hence specified traffic carrying circuits entering a TAP can be cross-patched, usually automatically to outputs from the TAP, and can then occupy different time slots in different multiplex groups. Thus, the output signalling relationship can differ from the input signalling relationship.

Each TAP inter faces with the Management System, which determines onward routing of signalling mini-packets from the incoming routing octets, and'the physical link identity and signalling relationship.

At the input, TAP 1, System Management holds connectivity information from which the onward routing of both the traffic carrying circuits and the related CCS messages can be determined from the physical identity of their destination. This is of the link over which they arrive, plus the identity, obtained by Layer 2 message analysis, of their destination in terms of LAP Address (for DASS-2/DPNSS), or TEI (for I.4xx) as appropriate. The references to layers relate to the well known seven layer protocol (OSI).

At intermediate points, TAPs 2 and 3, System Management uses its connectivity information to route onward both the traffic carrying circuits ane the related signalling messages according to the identity of the link and multiplexed group signalling relationship over which they arrived and the information in the routing octets, by substituting new routing octet(s) for the original octet(s), and, possibly, due to circuit grooming and consolidation generating a new signal relationship for the output to the next inter-TAP link.

At the output, TAP 4, System Management uses its connectivity information to output both the traffic carrying circuits and the relate signalling messages according to the identity of the link and the multiplexed group signalling relationship and the information in the routing octets. As the signalling is output to the peripheral, the routing octets are stripped off, and replaced b such Flag sequences as are needed by the CCS message signalling systems to the peripherals. For example, for DASS-2/DPNSS output signalling, at least one Flag is inserted between consecutive messages: additional Flags are introduced where needed to provide a "fill" in the absence of a signalling message. In the I.4xx case, there is a Flag sequence at the commencement and end of the signalling message: wnere there is no signalling message, there is no transmission.

Where a signalling message occupies two or more mini-packet, it is segmented at the input TAP into more than one mini-packet. After routing across the network via the intermediate TAPs as individual mini-packets, analysis of the routing header at the output TAP causes the mini-packets to be stored until all have been received, after which the routing headers are stripped off and the original messages are reconstructed prior to transmission to the peripheral as a single frame message.

From the above, it is clear that great reliance is placed upon the System Management subsystem for routing both the traffic carrying circuits and their associated signalling, particularly if it is of the message based form.

Tnus System Management contains information as to the network connectivity and peripherals so that the routes to be taken through the network can be defined and set up. Further, selecting of the routing octets to define the signalling relationship fora traffic carrying channel and its associated message based signalling common channel is a responsibility of the system management in the local TAP together with the control elements in the TAP.

System Management thus covers much more than the more traditional role of service quality and fault reporting. While it still does this, it also contains the intelligence to define and set up both direct routings and alternative routings to circumvent faulty facilities or ones giving poor quality of service, e.g.

increased error rates.

To provide the communications needed to support System Management throughout the network, a CCITT X.25 facility may be dedicated for this purpose throughout the whole network.

Thus by using routing octets "wrapped" n around the existing message based signalling, a flexible managed network between peripheral locations results.

While these are usually users (both subscriber groups and PABXs) and public, or private switching points the principle can be extended to provision of links between public concentrators and exchanges, and between public exchanges in the main network (e.g. junction facilities). Here it would be expected that CCITT No. 7 type message based signalling would be used, which may also be twrappedt with routing octets so that it can be carried through the flexible network.

Due to the "wrapping", the original signalling messages are preserved. Thus the access netWork is transparent, and a single LAP exists between any pair of peripherals, albeit with slightly increased transmission time due to the message analysis and possible segmentation at the entry TAP, minor processing at intermediate TAPs and reconstitution at the exit TAP.

Note that when using the mini-packets Flag sequences are only included where necessary to mark the end of a signalling message when that does not coincide with the end of a mini-packet. This is becausemini-packets are transmitted synchronously in relation to the multiplex group framing, so a Flag sequence to mark end of frame is not needed.

By transmitting idle mini-packets periodically, or at random, when there is no signalling, syncnronXsl., can De detected from tne receipt of the sequence of the 32 octets of a Mini-Packet without signalling content.

At the input TAP, full analysis of an incoming signalling message is still needed to determine its routing. This uses software to determine the addressing require and any segmentation. However, relatively simple hardware, rather than software, can be used to determine mini-packet synchronism, and to interpret the routing octets and substitute new routing octets at the intermediate TAPs. At the exit TAP, hardware recognition of the mini-packet routing octets causes the original message to be reconstituted, and to be transmitted to the appropriate peripheral.

Hardware rather than software processing may be more cost-effective where the processing is on a regular fixed cyclic basis, rather than processing by software when searching forsynchronisation, and extensive message analysis may be require.

Finally, the arrangements described permit routing message based signalling through a flexible managed access network, where the facilities of traffic circuit consolidation and grooming at intermediate nodal points is provided without needing to analyse the original message based signalling at each intermediate point.

For I.4xx signalling the SETUP" and some other messages contain the channel identification (i.e.

tne B channel identity) to which the message relates.

Note that a 144 Kb/s data stream includes A and B 64 Kb/s channels and other channels. For DASS-2/DPNSS signalling, layer 2 always contains the LAP identity, which indicates the B channel to which the signalling message relates Thus, if the grooming and consolidation causes signalling messages to output from TAPs to peripherals where the messages quote the original B channel identity, but in reality the B channel has been changed, the proposed schemes become unworkable.

Tnis is illustrated in Fig. 9 for I.4xx, and Fig. 10 for DASS-2/DPNSS. In both cases, at the output peripneral, the B channel used is at variance with the B channel identity in the signalling messages. A number of solutions can be provided to circumvent this situation.

First, grooming and consolidation could be inhibited. This preserves the B channel identity, but there is virtually no benefit from the use of such a network since flexibility is lost.

Second, grooming and consolidation could be allowed, provided that at the output TAPs, the original B channel identities were used for connection to the peripherals. This allows for flexibility strictlybetween TAPs, but flexibility in the connection to peripherals is constrained, defeating the objectives of a flexible network.

A third approach which overcomes the disadvantages of the other two is to change the B channel identity within the signalling messages such that at the output TAP the signalling message going forward to the peripheral contains the B channel identity to give the correct signalling relationship.

This permits full flexibility of consolidation, grooming, and re-routing. As this is the only way in wnich the full capaDilities of a flexible network may be realised, we now determine how this can be done without excessive additional slgnalling message analysis and processina.

At the output TAP, the output B channel identity within the multiplex group to the peripheral for a given connection is known fron the connectivity data in the Management System. Thus the B channel identity can De changed to the required value as part of the signalling message reconstitution at the output TAP.

Alternatively, it is possible to change the B channel identity within the DASS-2/DPNSS/I.4xx signalling messages at the input TAP where a greater degree of input signalling message analysis is performed to enable routing to be determined.

By substituting the outgoing B channel identity in the signalling messages at the input point, the Management System is relied on to provide the appropriate information. However, this is not a new requirement, rather a further use of the information for routing between peripherals. A direct consequence of substituting the B channel identify is that the original Frame CnecK Sequence needs to De changed if the substituted B channel identity differs from the original. This makes it necessary inthe input processing to examine the whole contents of the signalling message to perform the substitution.

With this level of input processing, two options become available.

First, the signalling message with the substituted B channel identity and the new FCS could be wrappedw with the routing octets and transmitted through the network to the output TAP, and thence, upon reconstitution to the peripheral. Here a single LAP exists between the input and output peripherals, in spite of the changes to the B channel identity and FCS value.

Second, the signalling message forming the LAP could terminate at the input TAP, and a second LAP be generated containing the message with the changed B channel identity and FCS value, which is then transmitted througn the network to the output TAP, and thence to the peripheral.

Tne second approach puts two LAPs in tandem, but has advantages. The general concept of message based signaliing of the types considered is that there is bi-directional flow of messages for most transactions, e.g. acknowledgements, retransmission requests, etc. Thus, with the use of two relatively independent LAPs as to realise such signalling across a flexible managed network, a basic requirement is that the LAPs in either direction relative to a pair of peripherals must terminate and interface at the same physical point in the access network. This could be at an intermediate TAP, or more likely, the input TAP for one direction, and the output TAP for the reverse direction.This is illustrated in Fig. 11 where the LAP for the "Input" peripheral, LAP (1) terminates at the "Input" TAP, and the LAP for the "Output" peripheral, LAP (2) also terminates at the "Input" TAP, where "Input" and "Output" are solely relative terms.

Again, note that LAP (1) from the "Input" peripheral terminates at the "Input" TAP, A and is also analysed for determining of the routing octets which are then "wrapped" around the output LAP, and LAP (2) containing the "Output" B channel identity substituted for the "Input" B channel identity. LAP (2) is then routed through via TAPs B and C, and reconstituted at TAP D prior to transmission to the "Output" peripheral.

By contrast, LAP (2) from the "Output" peripheral is only analysed at the "Output' TAP D from which the routing octets are determined and "wrapped" around LAP (2) to route it via TAPs C and D to the "Input" TAP A where it is terminated. The "Input" B channel identity must be substituted for the 'Output" B channel identity at this point.

This solves the problems of having different B channels for the inputs ane outputs, and ensures that the messages reflect the situation, and by fully terminating the LAPs, erroneous transmissions are not propagated. Further, TEI values, and, possibly, Call References (in the I.4xx case), can be assigned either on a fixes basis or by negotiation between the corresponding points within the LAPs.

Where one of a pair of communicating peripherals is at Basic Access, the LAP related to the Basic Access is terminated at the TAP with which it interfaces. At this point the second LAP is generated which is then routed to the other peripheral where it is terminated; there being no further intermediate LAP terminations.

Claims (11)

1. A telecommunication system which inter connects a number of interface units each associated with one or more of a number of users of the system, wherein connections may be set up via the system between the users, wherein signalling relating to the connections to be set up via the system uses a plurality of different signalling protocols, wherein when a message using one of said protocpls arrives at a said interface unit that message is provided with a system header appropriate to the passage of that message through the system, and wherein when a said signalling message with a said system header reaches the interface unit appropriate to its destination the system header is removed from that message and the message is routed from the said interface unit to its destination, so that the system is fully transparent to that message.

2. A system as claimed in claim 1, wherein the interface units are connected serially together via links between the interface units so that a signalling message may have to pass through one or more of said interface units other than the two between it is to be sent, and wherein at the or each said intermediate interface unit traversed by a said message the system header is replaced by a new system header appropriate to the next interface unit to which that message is to be sent.

3. A system as claimed in claim 1 or 2, wherein within the system signalling messages are conveyed as fixed length packets of data, wherein if a message to be conveyed is too long to be conveyed in one of said fixed length packets it is conveyed in two or more said packets, and wherein the system headers for said packets include indications as to whether a packet is a single packet message or is part of a plural packet message.

4. A system as claimed in claim 1, 2, or 3, wherein the signalling protocols which are handled include DASS-2, DPNSS and I.4xx.

5. A system as claimed in claim 1, 2 or 3, wherein the signalling protocols to be handled include CCITT number 7.

6. A system as claimed in claim 1, 2, 3 or 4, and which includes a plurality of management networks each of which serves and controls a plurality of said interface units, and at least one further management network serving said plurality of management networks.

/. A telecommunication system substantially as described with reference to the accompanying drawings.

Amendments to the claims have been filed as follows 1. A telecommunication system which interconnects a number of interface units each associated with one or more of a number of users of the system, wherein connections may be set up via; the system between the users, wherein signalling relating to the . .

connections to be set up via the system uses a plurality of different signalling protocols, wherein when a message using one of said protocols arrives at a said interface unlt that message is provided with a system header appropriate to the passage of that message through the system, and wherein when a said signalling message with a said system header reaches the interface unit appropriate to its destination the system header is removed from that message and the message is routed fro the said interface unit to its destination, so that the system is fully transparent to that message.

2. A system as claimed in claim 1, wherein the interface units are connected serially together via links between the interface units so that a signalling message may have to pass through one or more of said interface units other than the two between it is to be sent, and wherein at the or each said intermediate interface unit traversed by a said message the system heaaer is replaced by a new system header appropriate to the next interface unit to which that message is to be sent.

3. A system as claimed in claim 1 or 2, wherein witnin the system signalling messages are conveyed as fixed length packets of data, wherein if a message to be conveyed is too long to be conveyed in one of said fixed length packets it is conveyed in two or more said packets, and wherein the system headers for said packets include indications as to whether a packet is a single packet message or is part of a plural packet message.

4. A system as claimed in claim 1, 2, or 3, wherein the signalling protocols which are handled include DASS-2, DPNSS and I.4xx.

5. A system as claimed in claim 1, 2 or 3, wherein the signalling protocols to be handled include CCITT number 7.

6. A system as claimed in claim 1, 2; 3 or A, and which includes a plurality of management networks each of which serves and controls a plurality of said interface units, and at least one further management network serving said plurality of management networks.

7. k telecommunication system substantially as described with reference to the accompanying drawings.

8. A telecommunication system which interconnects a number of interface units each associated with one or more of a number of users of the system, wherein connections between the users may be set up via the system, wherein signalling relating to the connections to be set up via the system may use a plurality of different signalling protocols, wherein when a message using one of said protocols arrives at a said interface unit that message is provided with a system header appropriate to the passage of that message through the system, wherein when a said signalling message with a said system header reaches the interface unit appropriate to its destination the system header is removed from that message and the message is routed from the said interface unit to its destination, so that the system is fully transparent to that signalling message, wherein within the system signalling messages are conveyed as fixed length packets of data, wherein if a signalling message to be conveyed is too long to be conveyed in one of said fixed length packets it is conveyed in two or more such packets, wherein if a signalling message to be conveyed is shorter than a said fixed length data that message is made up to the fixed length by a filler data such as flags, and wherein the system headers for said packets include indications as to whether each said packet is a single packet message or is part of a plural packet message.

9. A system as claimed in claim 8, wherein the fixed length packets each consist of 30 octets of data.

10. A system as claimed in claim 1, 2, 3., 4, 5, 6, 8 or 9, wherein error control in respect of signalling messages within the system relies on end-to-end error detection techniques.

11. A systems as claimed in claim 1, 2, 3, 4, 5, 6, 8, 9 or 10, wherein the links between the interface units are optical fibre links.