GB2027564A - Non-hierarchical telecommunication system - Google Patents

Abstract

Nodes T, to each of which one or more subscribers' sets are connected, are interconnected to form a non- hierarchical mesh of nodes between which communication takes place using digital transmission. When a node only serves one line, the nodal equipment is included in the subscriber's set. Each node has a microprocessor which assesses the route to be used for each call to be set up from or through the node, causing that message to be sent out via the appropriate link, or in the case of a terminating call to be sent to the node's own termination. This non-hierarchical system is relatively simple, and with the falling costs of electronics, more and more economical, and is more sabotage-free than is a hierarchical system. <IMAGE>

Description

SPECIFICATION Improvements in or relating to telecommunication systems This invention relates to automatic telecommunication systems of the decentralised type.

Contemporary telephone networks and switching systems are basically heirarchical in structure, which enables relatively simple routing strategies to be used at the switching nodes (i.e. the exchanges), and allows the switches to be in relatively large central offices. Hence the equipment is easily accessible lor maintenance, and expensive common equipment can be shared economically over what are in effect separate switching nodes co-located geographically for administrative convenience. However, there is the disadvantage from the security aspect that a large amount of equipment which is both important "system-wise" and expensive is concentrated at one location.

One consequence of the hierarchical structure of such a network is that the lines are basically star-connected, with calls routed into the node (= exchange) and out again. Hence the route taken by individual calls is longer than the minimum point-to-point connection, so that the amount of line plant needed is greater than would be needed for a direct connection system. A further disadvantage present in many hierarchical systems is that switches of different sizes and types are needed at different levels in the hierarchy, and in many cases at different stages within the exchanges.

An object of the invention is to provide an automatic telecommunication system in which the disadvantages referred to above are at least minimised.

According to the invention there is provided an automatic telecommunication system, which includes a number of nodes interconnected by links such that a relatively small number of links terminate at each of the system nodes, and one or a small number of system terminations connected to each of the nodes, thus providing an interconnected mesh arrangement in which each of the nodes is directly connected to nodes adjacent thereto in the system, wherein messages conveyed by the system in respect of connections set up therein are in digital form, wherein to set up a connection between terminations connected to two of the system nodes each message conveyed between those two nodes includes routing information so that each message for the connection passes from one of the terminations involved in that connection to the other termination involved in that connection via two or more of the system nodes, and wherein each of the nodes includes switching equipment whereby on reception of a message either that message is sent out via one of the links connected to that node, as required by the routing in information in the message, or that message is applied to a termination directly connected to itself, whereby the system is fully decentralised.

Such an arrangement is a practical realisation of an alternative to the known hierarchical systems, the alternative thus being an interconnected mesh arrangement in which each node is connected only to its nearest neighbours. In such an arrangement calls are routed across the network through as many nodes as necessary.

With the system nodes and terminals separated, we would have a two-level hierarchy, and if two or more terminals are connected to a single node the network at that node is still of the star-type. Hence with a completely non-hierarchical network, the nodes are co-located with the system terminals. In practice, this means that the switching equipment is in the subscribers instruments, or at least at the point at which such instruments connectto the outside line plant. Locating the switching at the subscriber's premises has the advantage that it eliminates exchange building costs, but has implications for system security. As the distribution of subscribers throughout the network is not uniform, some degree of star-connection may be necessary in the interest of economy.Frequently subscribers' stations exist along a line such as a valley or a street, and each node may then only have two neighbours each.

In such a non-hierarchical, or almost non-hierarchical, mesh-connected network it is necessary for sufficient intelligence to be provided at each node to route a transit call, i.e. a call other than one between adjacent nodes, to any terminal in the network. The transmission quality of the switches has to be adequate for multiple tandem operation, and the nodes and links have to be of adequate traffic capacity. In addition the equipment has to be reliable enough for remote operation and dispersed maintenance.

With current technological trends it is to be expected that the "hardware" needed to meet the above criteria will become progressively cheaper, so that any added cost due to dispersing the switching equipment should be more than compensated for by the saving in line plant cost.

Embodiments of the invention will now be described with reference to the accompanying drawings, in which Figures land 2 are diagrams explanatory of certain of the principles basic to the invention.

Figure 3 is a simplified schematic of a node used in a system embodying the invention, together with a terminal associated therewith.

Figures 4 to 8 are explanatory diagrams illustrating the interconnection of nodes in systems embodying the invention.

Figure 1 and its analysis are intended to be of assistance in establishing the approximate size ot a nodal switch for a system embodying the invention. Each node N has connected to it one terminal T and n alinks, n being 4 in Figure 1. Each terminal generates a erlangs of traffic, the average call distance is D, and the internode link distance is d. In this we have assumed, to simplify the discussion, that the links are of equal length.

The traffic a from each terminal T divides at its local node N into a/n erlangs per iink. Assuming that all calls are of average distance D, then traffic entering node N1 other than that due to its own terminal will not reach node NX at distance D from N1.

The a/n erlangs which reach N2 from N 1 divide in the ration at N2, but similar traffic reaches N2 over its other links. Hence a net transit traffic of a/n passes on towards NX, and locally-generated traffic from T2 also of a/n is added in, so that 2a 2aerlangs passes from N2 towards NX.

Between N1 and NX have D/a links, and the node at the start of each link adds a traffic of a/n to make a total of Da, Thus the unidirectional traffic per link is Da and the switched traffic at the node, including localterminal originated traffic is A A=a +D= a(1 + d) The values of D and dvary considerably, even within a single exchange area, but it is worth noting that the average call distance in the United Kingdom is about 7 km. Similarly the density of terminals, i.e.

subscribers' stations, varies between 0.1 and 100 subscribers per hectare, the commonest density being around 5 subscribers per hectare, which gives a value for of 0.04 km.

For a subscriber-originated traffic of a = 0.05 E, traffic for a rectangular mesh (n=4) becomes (see Figure 2): Da 7 x 2.1875 (each way (i) perlink dn 0 04 x =2.1875 (each way (ii) pernode a(1 +D =005(1+) = 8.8E (total switched).

d 0.04 As will be seen from Figures 4 to 8, alternative mesh structures exist, and by suitable choice of mesh structures a variety of traffic patterns can be met by a single design of node.

We now consider the nodal hardware, Figure 3, in which the upper boxed area is the terminal while the lower boxed area is the node. It is here assumed that the terminal is a speech-only instrument having a key pad 1, microphone 2 and receiver 3. However, the system may include other sorts of terminal. The terminal shown has four incoming links and four outgoing links.

Since the system uses digital transmission, it operates on a four-wire basis, so that the hybrid usually needed in a conventional system is not needed. The keypad 1 interfaces directly with the nodal electronics, as seen by its connection to the scan circuit 4 via which it has access to a processor 5 with its associated memory 6. The processor 5 can be either a microprocessor suitably programmed or a special purpose processor. The memory 6 is preferably a semiconductor memory as such memories which are very small and have a very high capacity are available.

When a call is initiated, the number keyed in at the keypad 1 passes via the scan circuit 4 at a time in its cycle appropriate thereto to the processor 5. This determines which of the four outgoing links is to be used for the call, and via the Send box 7 causes the appropriate routing information extracted from the memory 6 to be sent over the link to be used for the call. The call initiating-message passes via one or more links to the "called" node, where the wanted terminal, if free is seized for the connection and an appropriate acknowledgement message sent back. if the wanted terminal is busy a busy code is returned to the calling node and therefrom to the terminal.

When the connection is "set up" as described above the switch 8 is set so that speech from the microphone 2 is coded in a coder 9 and sent via the box 7 to the wanted line. Incoming speech from the called line passes via the switch 8 to a decoder 10 from which it is applied to the receiver 3.

The scan unit 4 scans information incoming to the node over the four links and causes any messages therein for other nodes to be routed via the switch 8 and send unit 7 to the outgoing link appropriate to the messages' destination. This takes place under the control of information extracted from the memory 6. If the message is for its own terminal, scan 4 causes seizure under control of the processor 5 if it is the first message for a call. If the terminal is busy, scan 4 causes the sending back (over the RETURN path as it is a "four-wire" system) to the caller of a busy signal as mentioned above. Subsequent messages are routed as just described, and return messages from the called to the calling line are sent back over the return path in the same way as are messages from the calling to the called line over the forward path.

In view of the relation between the terminal and the node, multi-frequency signalling equipment as usually used with push-button sets is not needed. The absence of these components and of the hybrid usually needed makes room in the sub-set for the various electronic units shown in Figure 3.

As we do not have a central exchange power cannot be fed over the lines in the usual manner. Hence it is preferred to replace the usual carbon microphone by some other form of microphone, e.g. a moving-coil microphone. Power for the electronics can be derived from the mains where available, preferably with small rechargeable standby batteries. Where there are terminals at which mains is not available locally, power can be fed as direct current over the network links, e.g. phantom-wise on the two pairs, from the nodes at which mains is available. It would also be possible for the power supply to be by the use of special power-feeding nodes distributed throughout the system.

The scanning, coding, decoding and sending may be most economically dealt with by dedicated logic rather than under processor software control.

As will be seen later, the essential function of the memory 6 is to hold data representative of the current state of the network as "seen" by its associated terminal. The processor 5 determines from the identity of a wanted line and the data in the memory the routing information for the call. It will be appreciated that with a system such as the present, alternative routes will exist for most calls.

The swtich block 8 operates in time-divided manner, and routes each message under processor control to the local terminal or to a free time slot on one of the multiplexed external links. Thus we have both time and space switching, and the channel selection algorithm may need to be chosen carefully to minimise the number of time slot changes to avoid the introduction of excessive delay distortion on multiple tandem connections.

The terminal also has an indicator 11 which is enabled to provide either audible and/or visual calling signal when an incoming call is received.

At present it would appear that the most "cost-effective" solution for a system such as described above in which the nodes are made using LSI is likely to be "hardware-orientated", it may be useful to use software even for invariable functions since this would facilitate device development, system testing in service, and adaptation for various applications.

The organisation of the data in the memories such as 6, Figure 3, can be effected in a number of ways. The preferred method is to identify each memory address directly with a directory number. The data stored at each such address then only needs to be sufficient to enable selection of the exit routes for calls to the appropriate number. The assignment of memory addresses in directory number order means that the conversion of received digits into address is a simple arithmetic function. The data stored for each number can also include alternative routing data, useful when the system suffers from congestion. In such case the stored data includes route length digits so that for a call to be set up the processor chooses the shortest available route.

The generation of the route length digit referred to above is inherent in the updating of data. Thus when a new node is cut into service, it sends its directory number (or numbers if it serves more than one terminal) over all its exit routes with a route length digit of zero. Each node which receives this transmission increments the route length digit by unity and passes the message on. Each node also stores the route length digit against the appropriate directory number.

Testing and diagnostic functions consist in the main of the interchange of data and test messages between each terminal and its neighbours. Diagnosis of faults down to detailed hardware items is unnecessary if the nodal electronics is based on a single integrated circuit.

Security against component failure is less onerous than with a centralised exchange since its effects are usually limited to one subscriber. To prevent fault conditions from propagating through the system, each node is programmed to reject unwanted messages, and has protection against damaging line voltages.

Although the switching equipment is more accessible to tempering, this is less likely as some degree of knowledge is needed to be able to usefully interfere with the system.

Fraudulent calls are possible for anyone with the knowledge and skill to access his own terminal's memory, or to insert an unauthorised node into the system. However, the regular interchange of test messages as a background activity by all nodes enables the detection of the momentary interruption of the circuit due to such illegal activities. Analysis of traffic enables the identification of the origin of calls which can then be checked against records and charges to detect unmetered calls and unauthorised additions to the network.

Physical sabotage, or accidental damage, is hardly likely to effect overall system operation due to the extremely high degree of decentralisation. However, software based sabotage is a possibility, but it is easy to detect and to counter by the transmission of correcting data.

For administrative purposes, one or more special nodes used for operations and maintenance can be included in the network. Such nodes interrogate other nodes to extract stored data for metering and traffic recording, to advise all nodes of ceased numbers, and to make other program changes needed, subject to security precaution. At least one of these administrative nodes has a secure memory containing a record of the current state of the network and of authorised changes. It countermands any message detected as trying to make unauthorised changes.

It may be desired to dedicate one channel in each link to ensure that administrative messages can always reach every terminal so that it is not possible to isolate a section of the network due to real or false traffic. An alternative to this is for each node to be programmed to give absolute priority to administrative messages, even if this necessitates breaking down other calls. A similar "rig ht-of-way" treatment is used for emergency service calls.

Special services, such as operator position, are also allotted special nodes distributed as needed throughout the network.

The system has so far been described as a uniform mesh, e.g. as in Figure 4, where each ring is a node and each letter T is a termination. This arrangement may introduce some delays due to setting up a connection via several links, and this may be minimized by the use of some longer links, as can be seen in Figure 5.

Where such longer links exist, the route length digits referred to above would automatically route calls over such "non-adjacent" links.

To limit data storage needed per node, the fully interconnected network may extend only over a defined area corresponding to the service area of a conventional central exchange. The equivalent of junctions to adjacent areas are provided from nodes allocated for that purpose and distributed throughout the area, as called for by traffic needs. Thus in Figure 6, which is a simplification of a network representing subscribers along both sides of two roads, we have one node J N for junctions. Here we show one (TDM of course) junction J. The termination marked A is one of the nodes referred to above which are used for administrative purposes.

Figure 7 is similar to Figure 6 except that the uppermost links are replaced by one fully multiplexed link.

Figure 8 shows a network in which nodes are situated at a street distribution point to form a street concentrator.

The latter two systems both provide concentration of subscriber traffic which saves pairs in the cable used in a conventional system for connection to a central exchange. Such saved pairs can either be recovered or re-allocated to form extensions of the mesh for new subscribers.

Claims (7)

1. An automatic telecommunication system, which includes a number of nodes interconnected by links such that a relatively small number of links terminate at each of the system nodes, and one or a small number of system terminations connected to each of the nodes, thus providing an interconnected mesh arrangement in which each of the nodes is directly connected to nodes adjacent thereto in the system, wherein messages conveyed by the system in respect of connections set up therein are in digital form, wherein to set up a connection between terminations connected to two of the system nodes at least the first message between those two nodes includes routing information so that each message for the connection passes from one of the terminations involved in that connection to the other termination involved in that connection via two or more of the system nodes, and wherein each of the nodes includes switching equipment whereby on reception of a message either that message is sent out via one of the links connected to that node, as required by the routing informaton in the message, or that message is applied to a termination directly connected to itself, whereby the system is fully decentralised.

2. A system as claimed in claim 1, wherein when only one termination is connected to each of the nodes, that node and that termination are co-located so that the switching equipment individual thereto is at the subscriber's instrument.

3. A system as claimed in claim 1 or 2, wherein the switching equipment in a said node includes a switch via which an incoming call is routed to a local termination or incoming and outgoing links may be interconnected as called for by the routing information, a processor which controls said switch, and a memory from which inform as to the condition of the system is extracted to deal with each said message.

4. A system as claimed in claim 3, wherein a keyset at the termination from which wanted numbers are sent is connected at the node to a scan circuit under whose control wanted number information is conveyed to the processor, and wherein said scan circuit also scans all incoming messages to determine whether they are for a termination connected to the node.

5. A system as claimed in claim 3 or 4, wherein speech or other message information is coded at the node into digital form for transmission, each such message having said routing information added to it under control of the processor, and wherein messages received at the node for a termination connected thereto are decoded at the node so that communication between a termination and a node are in analogue form whereas communication elsewhere in the system is in digital form.

6. A system as claimed in claim 1, 2,3,4 or 5, in which the intelligence transmitted includes route length digits each of which is derived by the transmisson for a node in respect of which such a digit is needed a message in which the route length digit is incremented at each node via which a message is said for a new station.

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