WO2001017151A1

WO2001017151A1 - Dual protection arrangement for fdm rings

Info

Publication number: WO2001017151A1
Application number: PCT/CA2000/000983
Authority: WO
Inventors: Dan Parsons; John Boyd; Charles Choo; Steve Knobel
Original assignee: Impath Networks Inc.
Priority date: 1999-08-26
Filing date: 2000-08-24
Publication date: 2001-03-08
Also published as: AU6678700A

Abstract

A communication system includes a head end, a plurality of nodes, and a pair of transmission lines forming respective rings for carrying signals in opposite directions between said nodes to provide full duplex operation. A switch is located at each node for selectively breaking the continuity of each ring. The switch is normally open and the switches at the remaining said nodes are normally closed so as to divide the rings into two dynamically changeable logical segments. In the envent of a line break, the normally open switch closes, and one of the other switches between the break and the head end opens so that the dynamically changeable segments terminate on either side of the break. Node communications terminate at the head end. Node communication services switch direction of transmission to the head end based on a determination of a valid transmission path to the head end.

Description

DUAL PROTECTION ARRANGEMENT FOR FDM RINGS

This invention relates to ring networks suitable for carrying signals, such as video signals, and high and low speed data, and in particular to fiber optic ring networks employing frequency division multiplexing to carry the signals. Fiber optic rings are used in digital applications to carry high speed data. For example, SONET (Synchronous Optical Network) is a physical layer standard for digital transmission at speeds of up to 2.48 Gbits/sec.

Fiber optic lines are subject to breaks, for example, as a result of a mechanical digger inadvertently breaking into a line, and as a result it is common practice in SONET networks to provide counter-rotating rings so that bi-directional data is transmitted to each node in opposite directions. If a break occurs anywhere in one ring, the nodes will still receive the data from the other ring.

Digital technology is still complex and expensive to implement. Typical ring networks require a complex communication channel to manage the ring. There are many applications where it is preferable to employ analog technology. For example, video from security cameras placed at a site can be more economically sent in analog form to a head end than in digital form. Multiple video channels can be more economically multiplexed onto an optical fiber using frequency division multiplexing (FDM) than individually digitized and digitally multiplexed. Protection of FDM rings against line breakage is a problem. The techniques employed for digital rings cannot be utilized because for full duplex operation the attached devices must be able to use the same frequency to transmit and receive data in order to provide maximum bandwidth. The same path cannot be used for transmission and reception since collision of the frequency channels leads to signal corruption. An object of the invention is to provide an FDM ring with protection against line breaks.

According to the present invention there is provided an analog communication system comprising a head end, a plurality of nodes, at least one pair of transmission lines capable forming respective rings for carrying signals in opposite directions between said nodes to provide full duplex operation and, switch means located at each node for interrupting the continuity of each ring at each node, the switch means at one said node being normally open and the switch means at the remaining said nodes being normally closed so as to divide the rings into two logical dynamically changeable segments, and in the event of a line break, said normally open switch means being operable to close, and one of said normally closed switch means adjacent said break being operable to open so that said dynamically changeable segments terminate on either side of said break.

Node communications terminate at the head end. Node communication services switch direction of transmission to the head end based on a determination of a valid transmission path to the head end.

The invention employs an open loop architecture wherein the rings are divided into dynamically variable logical segments to prevent collisions between channels. The segments initially terminate at an arbitrary node designated the master node, but in the event of a break the segments change dynamically so as to terminate on either side of the break. If a break occurs between the node adjacent the head end and the head end, then there is only one active segment since all the nodes communicate along it, but the stub segment attached to the head end is still logically considered a segment.

The individual nodes have add/pass/access capability. Each node typically includes two channels, one for each ring direction. Incoming signals are passed through a repeater to the outgoing link except through the node which is normally open. Initially service channels are established with the head end in a predetermined direction depending on the segment. When a break occurs, this is detected by the node upstream of the break. This node then puts its repeaters into open circuit, and the repeaters in the original node close so as to complete a bi-directional path to the head end in the direction away from the break. The nodes upstream of the break switch direction so as to receive and transmit in opposite directions. The transmission lines are normally optical fibers, for example, carrying frequency division multiplexed video signals. In this case, a break can easily be detected by monitoring the optical level of the incoming signal. However, the transmission lines can also be electrical transmission lines, for example, co-axial cables.

In a preferred embodiment, a looped signaling channel is established in ring or in each segment. In the event of a break, this enables the nodes to determine whether a clear path exists back to the head end. The signal channel may carry a pilot tone, for example at 4MHz.

An advantage of the invention is that a dedicated communication control channel for inter node communication is not required. Such a control channel typically requires extensive processing.

The invention also provides a method of protecting a network analog communication system comprising a head end and a plurality of nodes, comprising the steps of providing at least one pair of transmission lines capable forming respective rings for carrying signals in opposite directions to provide full duplex operation between said nodes; interrupting said rings at an intermediate node to form two dynamically changeable logical ring segments, each providing full duplex communication to said head end; monitoring said rings at each node to detect a break therein; and upon detection of a break in said rings at another node, closing said rings at said intermediate node and breaking said rings at said another node so that said dynamically changeable segments terminate on either side of said break.

The invention will now be described in more detail, by way of example only, with reference to the accompanying drawings, in which:-

Figure 1 shows a possible network topology for a ring network;

Figure 2 shows the collision of frequency channels in a dual ring topology; Figure 3 illustrates a dual ring FDM network offering full duplex communication;

Figures 4a and 4b show an open ring architecture;

Figures 5a and 5b show the ring configuration in the event of a break;

Figure 6 shows a multiple ring architecture;

Figure 7 is a block diagram of a node including a tone generator; Figure 8 shows the presence of the pilot tone on the ring;

Figure 9 shows the default an alternate paths to the head end 12;

Figure 10 shows the path of the pilot tone through the head end; and

Figure 11 shows a second embodiment of the invention. In Figure 1, a plurality of nodes 10, which may be T1500 multiplexers from z^'MPath Networks Inc., are connected to a head end unit 12, which may be a T5000/2500 multiplexer, also from /MPath Networks Inc. by fiber optic rings 100, 200. Each ring carries frequency division multiplexed signals, which may, for example, be video and low speed data. The video and data signals can be added/passed/accessed at each node 10 in a manner known per se.

Typically, data is transmitted from the head end 12 to the nodes 10, and data from the nodes are 10 transmitted towards the head end. In a ring architecture, data transmitted from the head end, is received by the first node 10 and then passed to the next node 10 (i.e. the nodes are configured in Add/Pass configuration). This configuration allows all nodes to receive data, broadcast from the head end 12. Typically, as shown in Figure 1, a simplex channel may be sufficient for high speed data, such as video, where it is desired to collect video signals from a number of sources and send them to the head end 12. The low speed data is usually duplex since it is often desirable to send such data in both directions. However, duplex high speed channels can also be established if desired, for example, for video conferencing.

In order to maximize bandwidth, it is desirable to use only one frequency for full duplex communication with each node, i.e. for both transmission and reception. Consequently, the same frequency cannot be used to send as well as receive data from the head end on the same ring since the transmitted data will collide with the received data. This is shown in Figure 2.

In Figure 2, each node 10 receives an incoming frequency division multiplexed signal and drops part of the signal in splitter 101 to the local node. The closed repeater switch 102 forwards the incoming signal to combiner 103, where it is combined with a local generated signal to be added to the ring. It will be seen that if the received FDM signal has a component B at frequency f2, and the added signal has a component W at the same frequency, the outgoing signal will be corrupted due to a collision of the two frequencies. As a result separate paths are needed for full duplex communication as shown in Figure 3, where the data travels in opposite directions on rings 100, 200. Each node receives on one ring and transmits on the other in the direction toward the source of the received data. As shown in Figure 3, ring 100 permits communication in the anti-clockwise direction and ring 200 permits communication in the clockwise direction on ring 2. At each node 10, ring 200 contains splitter201, repeater 202, and combiner 203. In Figure 3, data from the head end node 12 is broadcast to nodes 10A, 10B, 10D using frequencies fl, f2 and f4 respectively. Node 10D, for example, transmits on frequency f4 to the head end 12, but receives on ring 200 so that no collisions occur with the data on ring 100.

Thus, for example, node 10 transmits data to the head end 12 anti-clockwise around ring 200 on frequency f4 and receives data coming clockwise from the head end around ring 100 also on frequency f4. As the node transmits and receives on different rings, no collisions occur. However, the head end does not need to receive data on ring 100 as it is the same as it sent out originally, and furthermore, it must not send out data on ring 200 since such data would collide with data transmitted from the individual nodes. In effect, the rings must be interrupted in order to prevent frequency collisions.

Figure 4a shows a practical embodiment of the invention in which the rings 100, 220 are intentionally interrupted at center or master node 1 OC. This is achieved by opening the repeater switch 102, 202 for each ring 100 in the node IOC. The result is that the network is divided into two segments 120, 220 as shown in Figure 4b. Now the data flow to the head end 12 is shown by the arrows on the rings. On segment 120, data flows from the head end 12 to nodes 10A, B, C on ring 100 and in the reverse direction on ring 200. On segment 220, data flows out on ring 200 through nodes 10E, 10D and back on ring 100 through nodes 10D, 10E. The head end 12 can still broadcast to all nodes in both directions and forward received spectrum without the risk of collision because of the break at node IOC.

It is possible for the ring to be configured so that it is intentionally interrupted at any node, which then serves as the master node, including at node 10E, in which case the segment between node 10E and the head end 12 is merely a logical segment that in the normal condition in the absence of breaks carries no data. All communication would take place with the head end via nodes 10D, 10C, 10B, and 10A under normal conditions. In the event of a break, the logical segments would reconfigure in a manner to be described. In the event of an unintentional break in the rings, repeaters in node 10C close, and the repeaters in the node immediately upstream of the break in the receive direction open to interrupt both rings at that node. This event is shown in Figure 5a, where a break occurs between nodes 10A and 10B in ring 200. The node upstream of the break in the receive direction, node 10 A, detects the break and opens its repeaters in 102, 202 in each ring 100, 200 to break the transmission path. At the same time, repeaters in node 10C close so that the two ring segments now terminate at node 10A. Now nodes 10B, C, D, E communicate with the head end 12 over new segment 220' while node 10A communicates over new segment 120', i.e. directly with the head end 12. Node 10B starts receiving data from the head end 12 over ring 200 instead of ring 100. No frequency collisions occur because the signals are now interrupted at node 10A in both directions. Head end 12 can transmit to node 10B on ring 200, for example on frequency £2, and node 10B can transmit on frequency f2 on ring 100. The £2 signal from the head end to node 10B on ring 100 does not collide with the transmitted signal because of the new break at node A. The advantages of the dual ring architecture are maintained and communications are ensured in the event of a break. When the break 300 has been repaired, node 10C again opens and node 10A closes, thus causing the system to revert to its original preferred configuration, whereby communication with node 10B takes place with the head end 12 via node 10A.

If a second break occurs on each ring between the same pair of nodes 10, the architecture allows communications between all nodes 10 and the head end 12 to be maintained. However, if several breaks occur between different nodes 10 at the same time, the intervening nodes 10 will be out of service. For this reason, it is important that optic fibers in close proximity to each other run between the same pair of nodes since these are likely to be taken out together.

An advantage of the open ring architecture, which essentially divides the ring into two segments, is that it lends itself to multiple ring architectures. Such a multiple ring architecture is shown in Figure 6.

In Figure 6, a master dual ring 400 connects head end units 412, which may be T5000 transporter units by z^'MPath Networks Inc. A logical break occurs at 401 dividing the master ring into two separate logical segments. Each head end 412 forming part of the master ring 400 also serves as the head end for respective subordinate dual rings 420 A, 420B, 420C containing nodes 410 and having logical breaks 402, also dividing them into pairs of segments 430, 440. The break points X, which ensure no collisions occur, can be moved around the dual rings when a break occurs in the same manner as discussed above. For example, under normal conditions, node 410A will communicate with head end 412A via its own head end over segment 430 of ring 420 A and via segment 450 of the master ring 450. If a break occurs at Y, the break 402 at X will be closed, a break will be created at Y between node 410D and 410, so that node 410 will then communicate over segment 440 with head end 412B. The communication between head end 412 B and head end 412A will occur in the same manner as before. Similar reconfiguration of the segments can occur in the master ring 400. In order to permit an upstream node to determine when a break has occurred and when it has been repaired, a signal, for example, a pilot tone, typically of 4MHz is originated at one node, for example, but not necessarily the head end 12, and transmitted completely around each ring 100, 200 in one embodiment. In the normal state, in the absence of a break, the pilot tone is forwarded by all nodes, including the master node 10C, in both directions regardless of whether the repeater is in the open or closed state. In a preferred embodiment described below, the pilot signal is transmitted around each of the respective segments.

Each node 10 continually monitors the incoming signal for the pilot tone on each receive channel. Of course, each node has two receive channels and two transmit channels, one each for each ring. As long as a tone is received on each receive channel, the node knows that there are no physical breaks in the ring. The master node 10C forwards the pilot tone regardless of the state of its repeaters.

Under the conditions shown in Figure 5a, when the physical break 300 occurs, node 10A immediately observes a loss in optical power on the receive side for ring 200, indicating a physical break in the ring. It then stops repeating the pilot tone in both directions. Nodes 10B stop receiving the pilot tone on the receive side on ring 100, but continue to receive the tone from the head end on the receive side on ring 200. This indicates that there is no longer a complete path to the head end 12 in both directions over segment 120.

In Figure 7, which shows the configuration of a node in more detail, the node 10 is based on a T1500 node sold by zMPath Networks Inc. It includes optical transceiver pairs 510, 520 connected to respective optic fiber rings 100, 200. Each transceiver pair consists of a transmitter on one ring and a receiver on the other so as to provide full duplex capability in each direction. In each path, as described with reference to Figure 3, there is a splitter 101, 201, a repeater switch 102, 202, and a combiner 103, 203. An optical level detector 105, 205 at the input of each ring monitors the level of the incoming energy. The break 300 is initially detected by the optical level detector 105, 205, which identifies a physical break upstream of the node when no signal is detected at the receive input of the corresponding transceiver. Upon detection of the break, the detecting node informs all downstream nodes in its segment by failing to repeat the 4MHz pilot tone on both transmitters. Thus, for example, in the case of Figure 5b, the level detector 205 for node 10A observes a loss of signal and thus identifies a break upstream of the node on the associated ring. This node 10A then stops forwarding the pilot tone in both directions, thereby informing the head end 12 and nodes 10B and 10B that a break has occurred. Upon detection of the loss of pilot tone, the center node 10C closes its repeater switches to complete the path through it on each ring. As shown in Figure 5b, the segments change dynamically to terminate on either side of the break. When a node detects a loss of pilot tone on the receive portion of the transceiver it is using to communicate with the head end, it switches to the other transceiver to communicate with the head end over the other segment, which dynamically changes to incorporate the node that was formerly in the other segment. In Figure 7, a 4MHz detector is selectively connectable to splitters 101, 201 by a selector switch 108 to receive the incoming signal and detect the presence of a pilot tone on the receive side of either ring to confirm that no break occurs between the current node and the node immediately upstream.

Similarly, a 4MHz tone generator 110 is connectable through switches 111, 112 to the combiners 103, 203 to insert the 4Mhz pilot tone on the outgoing ring segments.

Each node, regardless of its condition, transmits and monitors a pilot tone so that the absence of such a tone will always indicate a break upstream of the node. The transmission of the pilot tone is shown in Figure 8.

In normal operation, for each node there are possibly two paths that it can use to transmit towards the head end. Each node is preconfigured such that one of its transceivers pairs 510, 520 is designated as the optimum transceiver to use in order to communicate with the head end. This transceiver is known as the default transceiver. The other transceiver is referred to as the alternate transceiver. The optimum path to use is determined by the least number of hops the node is from the head end. This is shown in Figure 9. If the node 10 has both paths available to transmit towards the head end 12, it will use the default transceiver. A node 10 provides downstream nodes a path to the head end by enabling both its repeaters 102, 202.

By assigning the default path to use, based upon the number of hops towards the head end 12, it can be seen that the ring of nodes 10 is then divided into the two logical segments. Each segment contains nodes that are communicating in the same direction towards the head end. This naturally determines the point in which the ring is to be opened when there are two complete paths towards the head end, as shown in Figure 8. The ring must be opened when there are two complete paths to the head end, in order to avoid collisions as discussed earlier . The ring is opened by pre-assigning a particular node to open its repeater switches when it detects that the two paths are available. In a ring with an odd number of nodes, this would normally be the node furthest from the head end 12. This node can be configured to have its default path on either transceiver, as both paths contain the same number of hops towards the head end. Given an even number of nodes, there will be two nodes that will be equidistance away from the head end. One of the nodes is configured to open the ring when both paths are available. The node, which is assigned to open the ring (disabling its repeater function) when it detects both paths to the head end is not obstructed, is usually known as the Center Node.

As discussed above, the 4MHz pilot tone, within the RF spectrum, is used to indicate to each node that a path to the head end is unobstructed. The 4MHz signal is originated at the T5000/T2500 forming the head end. The head end transmits the 4MHz signal on both rings, since there are two paths to the head end. A node receiving this signal on an transceiver knows that it can access the head end using that particular transceiver. Each node 10 receiving this signal, repeats the signal to the next node to indicate to it that the path is clear. In a ring where there are clear paths in both directions to the head end, the Center Node also repeats this 4MHz signal, even though its signal path repeater function is disabled. Figure 8 depicts the 4MHz generation in a ring in which there are no obstructions of the paths in both directions.

When both paths are not obstructed and the Center Node forces a break in the ring, the Center Node then becomes the tail end of one segment and the adjacent node becomes the tail end of the other segment as seen in Figure 9. When there is a physical break in the fiber, the tail ends must move to the nodes at either side of the break as shown in figure 5b. The node whose optical receiver link has been broken disables its repeater function on both rings, in order to avoid collisions if only one ring is physically broken. The node which detects the break, must inform all nodes on either side of it, that they can no longer access the head end through this node. It does this by not repeating the 4MHz signal on both transceivers. Each node in turn will not repeat the 4MHz signal in order to inform other nodes that the path is obstructed. When a node detects the loss of this signal on the transceiver it is using to communicate with the head end, it will use the other transceiver to communicate with the head end, providing that it has an unobstructed path in that direction (i.e. it is receiving a 4 MHz clock from this direction). When the Center Node receives the loss of a 4MHz signal, it enables its repeaters in order to allow other nodes to communicate to the head end through it.

Once the break in the ring has been repaired, the node detecting the break (via the optical level detector) informs nodes on both sides of it that the path to the head end is available through it. It does this by repeating the 4MHz signal it receives. It then closes its RF switches in order to allow other nodes to communicate to the head end through it. The Center Node will detect that both paths are good and then will open its RF switches in order to open the ring to avoid collisions. The Center Node then becomes a tail end of a segment once again. Each node, detecting that both paths to the head end are available, will then use its default transceiver to communicate to the head end. Thus, the configuration of the is once again optimized for the least number of hops.

The head end 12, for example, a T5000/T2500 unit from zMPath Networks Inc, must provide a means to indicate to the attached nodes 10 that the optical link which they use to transmit to the head end is operational. Ideally, the head end should monitor the optical link from the nodes and transmit the 4MHz signal to the nodes 10 as long as the optical link is good. If the optical link is bad, the head end 12 should disable the 4MHz to signal this condition. Unfortunately, the optical detector in the head end 12 of a T5000/T2500 unit cannot indicate to the transmitter that the level is low. The end node adjacent the head end thus transmits a 4MHz signal to the optical receiver of the head end, which places it on its back plane. The optical transmitter is configured to obtain this clock from the backplane and send it out to the adjacent node. Therefore, if the optical link from the node to the optical receiver is broken, then the node will not receive the 4MHz clock which indicates to it that this path to the head end is unavailable. This configuration is depicted in Figure 10.

Because the nodes that are directly attached to the head end must constantly generate a 4MHz clock, a separate algorithm from the rest of the nodes must be used. The attached nodes are referred to as End Nodes. Also, if connected to an Optical Concentrator Card (OCC) the nodes are still be configured to constantly transmit a 4MHz signal to the head end in order that the End Nodes are not further divided into another algorithm.

An OCC does have the ability to generate a 4MHz clock to the attached T1500s, but it does not have the ability to stop transmission of the 4MHz signal to the node on a port basis when it detects a low optical level from the node. But since an OCC does have the ability to stop all RF transmission (which includes the 4MHz signal) to the node on a port basis, the OCC monitors the optical level and then disables all RF signals to the End Node if the optical level is low. The 4MHz signal transmitted from the OCC to the T1500 will be generated by the OCC onboard 4MHz clock. The following is the algorithm used to control the RF transmit switch of an OCC:

- generate 4MHz signal to End Node from on board 4MHz clock

- enable optical transmitter continuously if (received optical level is good)

{

- enable RF transmit switch

} else (optical level is not good) {

- disable RF transmit switch }

Rx service switch (default path 4MHz signal == good 4MHz signal) set Rx service switch to default path else if (alternate path 4MHz signal == good 4MHz signal) set Rx service switch to alternate path else r both 4 Hzpaths are bad */ set Rx service switch to default path

4MHz Tx switch (any optical level detector == bad optical link) disable both 4MHz Tx switch else if (both optical level detector == good optical links)

(default path 4MHz signal == good 4MHz signal) enable alternate path 4MHz Tx switch else disable alternate path 4MHz Tx switch

(alternate path 4MHz signal == good 4MHz signal) enable default path 4MHz Tx switch else disable default path 4MHz Tx switch

Figure 1 1 shows an alternative embodiment of the invention in the initial state. In the embodiment shown in Figure 1 1 , unlike the previous embodiment, there are only two types of nodes, namely slave nodes 10A, 10B, 10D, 10E and master or center node IOC. There are no END nodes as required in the above embodiment.

As in the earlier embodiments, the rings are divided into two dynamically changeable segments 120, 220. In this embodiment, the master node IOC transmits the pilot tone, in the form of a 4MHz clock signal, on each ring toward the head end on each segment 120, 220. The pilot tone is repeated at each node until it reaches the head end 12, which under normal conditions regenerates it out on the other ring of the same segment. For example, transmitter 400 of master node 10C transmits a pilot tone on in ring 200 to head end 12, which retransmits it on ring 100 via transmitter 402. Similarly, transmitter 403 of master node 10C transmits a pilot tone on ring 100 toward head end receiver 404. The head end receiver sends it back on ring 405. Either segment can service the master node IOC. In the normal state, the master node IOC is open so as to interrupt the ring. The slave nodes pass the pilot signal, which can be in the form of a network clock, and the transmitted data in both directions. The head end communicates with each node along the default segment as described above. When a node detects a loss of optical power in the receive direction, it stops repeating the clock signal in both directions. The remaining nodes in the segment then know that there is an incomplete path to the head end 12 in the receive direction on the ring where loss of signal has been observed by the node detecting the loss of signal. The master node IOC closes its repeaters and the node upstream of the break in the receive direction opens its repeaters in the manner shown in Figure 5b in connection with the first embodiment. All nodes will then switch their services to the direction from which they receive a valid clock.

For example, if the break occurs at X between nodes 10A and 10B on ring 200node 10B will detect loss of optical power. At this point it will open its repeaters and stop forwarding the pilot signal in both directions. As it lies on segment 1 , whereas in the normal condition it will communicate with the head end 12 by transmitting on ring 100 and receiving on ring 200, it will switch its services to transmit on ring 200 and receive on ring 100. At the same time master node 10 will close its repeaters so as to forward the data between the node 10B and the head end 12 over segment 220, which has now dynamically changed to terminate at node 10B instead of node IOC. Node 10B will continue to receive the network clock on the receive side of ring at receiver 406, but will of course not receive the network clock on the receive side of ring 200 at receiver 407.

When the break is repaired, node 10B will again start receiving the network clock and it receiver 407 on ring 200, which it will repeat to master node IOC. The latter will again open its repeaters, and node 10B will revert to communicating with the head end over segment 1 as normal conditions have now been restored. The embodiment is simpler to implement than the previous embodiment since it only requires two kinds of nodes apart from the head end, namely slave nodes and a master or center node.

The state of the system can be described as follows: Head end Initial State

Regenerate the received network clock from each segment back out to the network. Set which segment is to receive the network services.

Master Node Initial State Both transmission paths open.

Generate network clock to both segments.

Slave Node Initial State

Both transmission paths closed.

Regenerate (Pass) clock in both directions The following are the rules governing the switching conditions for a Master node:

The following are the rules governing the switching conditions for a Slave node:

The invention thus provides a bandwidth efficient method of providing communication protection in an analog environment, particularly in the case of frequency division multiplexed signal transmission without the overhead of an internodal communication link for ring management.

Claims

Claims:

1. An analog communication system comprising a head end, a plurality of nodes, at least one pair of transmission lines capable forming respective rings for carrying signals in opposite directions between said nodes to provide full duplex operation and, switch means located at each node for interrupting the continuity of each ring at each node, the switch means at one said node being normally open and the switch means at the remaining said nodes being normally closed so as to divide the rings into two logical dynamically changeable segments, and in the event of a line break, said normally open switch means being operable to close, and one of said normally closed switch means adjacent said break being operable to open so that said dynamically changeable segments terminate on either side of said break.

2. A communication system as claimed in claim 1, wherein said switch means comprises a repeater switch for duplicating an incoming signal on an outbound link.

3. A communication system as claimed in claim 2, wherein each node comprises a first channel for receiving and transmitting signals on a first of said pair of rings and a second channel for receiving and transmitting signals in an opposite direction on a second of said pair of rings, and a said repeater switch is provided in each channel.

4. A communication system as claimed in claim 3, further comprising a splitter on an upstream side of each repeater switch and a combiner on a downstream side of each repeater switch for respectively accessing signals from and adding signal to said channels.

5. A communication system as claimed in claim 3, further comprising means for generating a pilot signal on each ring, and wherein said nodes are arranged to forward said received pilot signal in the absence of a break in said rings.

6. A communication system as claimed in claim 5, wherein each said node comprises means for detecting the loss of signal on a ring in the receive direction, and switch means for preventing the transmission of said pilot signal in both directions upon detection of said loss of signal.

7. A communication system as claimed in claim 6 wherein said means for detecting loss of signal is a power level detector.

8. A communication system as claimed in claim 6, wherein said means for generating said pilot signal is located at the head end and said pilot signal is transmitted in opposite directions around said respective rings.

9. A communication system as claimed in claim 6, wherein means for generating said pilot signal is generated located at a node on each segment and transmitted around each segment.

10. A communication system as claimed in claim 9, wherein said means for generating said pilot signal is located at said one said node with said normally open switch means, and said pilot signal is transmitted around each said segment from said one node.

11. A communication system as claimed in claim 10, wherein said pilot signal is a radio frequency signal.

12. A communication system as claimed in claim 7, wherein said transmission lines are optic fibers, and said power level detector is an optical level detector.

13. A communication system as claimed in any one of claims 1 to 12, wherein said signals are carried on said transmission lines using frequency division multiplexing.

14. A communications system as claimed in any one of claims 1 to 13, wherein said pair of rings comprise a master dual ring and further comprise at least one sub-dual ring comprising pair of sub-ring rings connected to one or more of said nodes of said master ring, said at least one sub-ring having sub-nodes containing switch means operable in the same manner as said master dual ring.

15. A method of protecting a network analog communication system comprising a head end and a plurality of nodes, comprising the steps of:- providing at least one pair of transmission lines capable forming respective rings for carrying signals in opposite directions to provide full duplex operation between said nodes; interrupting said rings at an intermediate node to form two dynamically changeable logical ring segments, each providing full duplex communication to said head end; monitoring said rings at each node to detect a break therein; and upon detection of a break in said rings at another node, closing said rings at said intermediate node and breaking said rings at said another node so that said dynamically changeable segments terminate on either side of said break.

16. A method as claimed in claim 15, wherein in the absence of a break in a ring, a pilot signal is transmitted from said head end completely around each ring including said intermediate node, and in the presence of a break in a ring, the node upstream of the break in the receive direction ceases to repeat said pilot signal on each ring to notify the remaining nodes on its segment of the break.

17. A method as claimed in claim 15, wherein in the absence of a break in a ring, a pilot signal is transmitted completely around each segment including the head end, and in the presence of a break in a ring, the node upstream of the break in the receive direction ceases to repeat said pilot signal on each ring to notify the remaining nodes on the segment of the break.

18. A method as claimed in claim 17, wherein said pilot signal is generated at said intermediate node and transmitted around each segment.

19. A method as claimed in claim 18, wherein said signal is a radio frequency signal.

20. A method as claimed in claim 19, wherein said transmission lines are optic fibers.

21. A method as claimed in claim 17, wherein said rings are monitored with an optical level detector which indicates a break when the received optical signal falls below a predetermined threshold, and upon detection of said break by said level detector the associated node stops transmitting said pilot signal.

22. A method as claimed in claim 15, wherein said signals are frequency division multiplexed on said transmission lines.

23. A method as claimed in claim 15, wherein said pair of transmission lines provide a master dual ring, and one or more sub dual rings are connected to nodes of said master ring, each said sub dual ring being divided into ring segments in the same manner as said master ring.