WO2001052591A1

WO2001052591A1 - Telecommunications network distributed restoration methods and systems

Info

Publication number: WO2001052591A1
Application number: PCT/US2001/000449
Authority: WO
Inventors: Sig. H. Badt, Jr.
Original assignee: Alcatel
Priority date: 2000-01-07
Filing date: 2001-01-05
Publication date: 2001-07-19
Also published as: EP1159851A1; AU3441601A; JP2003520525A; CA2366373A1

Abstract

A network wherein a plurality of links connects a plurality of nodes such as cross-connects in a communication circuit network with paths interconnecting the nodes, and with there being spare capacity between a sufficient number of nodes to accommodate at least some rerouting of traffic immediately upon detection of a break in a traffic span in the network so as to restore circuit continuity within a predetermined maximum time using an improved failure detection, isolation, and recovery scheme.

Description

DESCRIPTION

TELECOMMUNICATIONS NETWORK DISTRIBUTED

RESTORATION METHODS AND SYSTEMS

BACKGROUND OF THE INVENTION

TECHNICAL FIELD OF THE INVENTION

The present invention relates generally to telecommunications systems and their methods of operation and for dynamically restoring communications traffic through a telecommunications network, and more particularly to a messaging method by which the origin and destination nodes of a failed path can receive information on which spans or links remain usable in the failed path.

This invention also relates to a distributed restoration algorithm (DRA) network, and more particularly to a method for isolating the location of a fault in the network and the apparatus for effecting the method, and more particularly to a method of monitoring the topology of the spare links in the network for rerouting traffic in the event that the traffic is disrupted due to a failure in one of the working links of the network.

This invention further relates to a distributed restoration method and system for restoring communications traffic flow in response to sensing a failure within spans of the telecommunications network, and even more specifically, to telecommunications network having 1633-SX broadband digital cross-connect switches.

DISCUSSION OF THE RELATED ART

Whether caused by a backhoe, an ice storm or a pack of hungry rodents, losing a span or bundle of communication channels such as DS3 and SONET telephone channels means losing significant revenues. After the first 1.5 seconds ofan outage, there is also a significant risk that the outage may disable one of more local offices in the network due to an excess of carrier group alarms. Several techniques are commonly used to restore telecommunications networks. Several of these are well known. The first of which is called route diversity. Route diversity addresses the situation of two cables running between a source and a destination, one cable may take a northward path, while the other takes a southward path. If the northward path fails, traffic may be sent over the southward path, or vice-versa. This is generally a very high quality restoration mechanism because of its speed. A problem with route diversity, however, is that, generally, it is very expensive to employ. The use of rings is another well-known technique that also provides for network restoration. This is particularly attractive when a large number of stations are connected together. The stations may be connected in a ring; thus, if any one connection of the ring fails, traffic may be routed in a direction other than the one including the failure, due to the circular nature of the ring. Thus, a ring may survive one cut and still be connected. A disadvantage with rings, is that the nodes of telecommunication networks must be connected in a circular manner. Without establishing the circular configuration that a ring requires, this type of restoration is not possible.

Another method of network restoration, mesh restoration, entails re-routing traffic through the network in any way possible. Thus, mesh restoration uses spare capacity in the network to re-route traffic over spare or under utilized connections. Mesh restoration generally provides the lowest quality of service in the sense that it generally requires a much longer time than does route diversity or ring restoration to restore communications. On the other hand, mesh restoration has the attraction of not requiring as much spare capacity as do route diversity or ring restoration. In performing network restoration using mesh restoration, two techniques are possible.

One is known as centralized restoration, the other is known as distributed restoration. In centralized mesh restoration, a central computer controls the entire process and all of the associated network elements. All of the network elements report to and are controlled by the central computer. The central computer ascertains the status of the network, calculates alternative paths and sends commands to the network elements to perform network restoration. In some ways, centralized mesh restoration is simpler than distributed mesh restoration. In distributed mesh restoration, there is no central computer controlling the entire process, instead, the network elements, specifically the cross- connects communicate among themselves sending messages back and forth to determine the optimum restoration path. Distributed mesh restoration, therefore, performs a level of parallel processing by which a single restoration program operates on many computers simultaneously.

Thus, while the computers associated with the network elements are geographically distributed, parallel processing still occurs. There is yet one set of instructions that runs on many machines that are working together to restore the network.

The telecommunications network is comprised of a plurality of nodes connected together by spans and links. These spans and links are for the most part fiber optical cables. A path is defined in the network as the connection between the two end nodes by which information or traffic can traverse. One of these end nodes is defined as an origin node while the other is the destination node. There could be a number of paths that connect the origin node to the destination node. And if one of those paths that carries the traffic is disrupted, another path can be used as an alternate for rerouting the traffic. Thus, if a path based approach for restoring disrupted traffic is employed in a telecommunications network and the network is provisioned with a distributed restoration algorithm (DRA), the origin and destination nodes are taken into consideration.

In the telecommunications network, it is likely that when a fault occurs, only a single link or span along a path is affected. Such a catastrophic event may be due to for example a cable cut. The remainder of the path through the network on either side of the failure may therefore remain intact and could be used for circumventing the failed portion. Yet a path based DRA provisioned network nonetheless may disregard the intact portions and seek a completely different alternate route for restoring the traffic. This can be a problem because indiscriminate selection of a candidate path to restore one failed path can preclude the restoration of other simultaneously, or subsequently, failed paths as damage to a single span in a network can cause the failure of many paths that may happen to pass through the same span. Therefore, a DRA provisioned network must take into account all such end to end paths that need to be restored in order to ensure the highest possible degree of restoration following a failure.

There is therefore a need for an improved DRA provisioned path based restoration scheme that allows the intact portions of a failed path to be ascertained and made available for forming alternate restoration paths, to thereby improve the efficiency and the completeness of the restoration. In a telecommunications network provisioned with a distributed restoration algorithm (DRA), the network is capable of restoring traffic that has been disrupted due to a fault or malfunction at a given location thereof. In such DRA provisioned network, or portions thereof which are known as domains, the nodes, or digital cross-connect switches, of the network are each equipped with the DRA algorithm and the associated hardware that allow each node to seek out an alternate route to reroute traffic that has been disrupted due to a malfunction or failure at one of the links or nodes of the network. Each of the nodes is interconnected, by means of spans that include working and spare links, to at least one other node. Thus, ordinarily each node is connected to an adjacent node by at least one working link and one spare link. It is by means of these links that messages, in addition to traffic signals, are transmitted to and received by the nodes.

In a DRA network, when a failure occurs at one of the working links, the traffic is rerouted by means of the spare links. Thus, to operate effectively, it is required that the spare links of the DRA network be functional at all times, or at the very least, the network has a preconceived notion of which spare links are functional and which are not. In addition to routing traffic, the links also provide to each node signals that inform the node of the operational status of the network. Thus, a signal is provided to each node to inform the node that traffic is being routed among the nodes effectively, or that there has been a malfunction somewhere in the network and that an alternate route or routes are required to reroute the disrupted traffic. Conventionally, when everything is operating correctly, an idle signal, or some other similar signal, is propagated among the various nodes of the network to inform those nodes that traffic is being routed correctly. However, if a fault occurs somewhere in the network that disrupts the flow of traffic, an alarm is sent out from the fault location and propagated to the nodes of the network. Such alarm signal causes the equipment in the network downstream of the fault location to go into alarm. To suppress the alarm in the downstream equipment, a follow-up signal is sent. This prior art method of sending out an alarm signal from the location of the fault aids in the fault isolation along a single link. Unfortunately, the standard that requires the sending ofan alarm signal downstream from the fault also requires that the downstream nodes, upon receipt of the alarm signal, further propagate it downstream. As a consequence, since all nodes in the network will receive the alarm signal within a short period of time after the fault has occurred, it becomes very difficult, if not downright impossible, for the management of the network to identify the custodial nodes of a failed link, or the site where the fault occurred. This is due to the fact that, in addition to the custodial nodes, many other nodes in the network likewise are in receipt of the alarm signal.

Therefore, a method is required a method in which the true custodial nodes of a failed link be made aware that they indeed are the custodial nodes. Putting it differently, a method is required to differentiate the alarm signal received by nodes other than the custodial nodes from the alarm signal received by the custodial nodes, in order to preserve the accepted practice of sending an alarm signal to downstream equipment.

Since in most instances a distributed restoration domain is a portion of an overall telecommunications network, or a number of different networks, it is therefore also required that the status of whatever signals received by the nodes outside of the distributed restoration domain be maintained as if there has not be any differentiation between the time when those signals are received by the custodial nodes and when those signals are received subsequently by the nodes outside the domain. There is also a need for the instant invention DRA network to always have an up-to-date map of the functional spare links, i.e. the spare capacity, of the network, so that traffic that is disrupted due to a failure can be readily restored.

SUMMARY OF THE INVENTION The present invention thus comprises the concept of connecting a plurality of nodes such as cross-connects in a communication circuit network with control channels interconnecting all nodes, and with there being spare capacity between a sufficient number of nodes to accommodate at least some rerouting of traffic as quickly as possible upon detection of a break in a traffic span in the network so as to restore circuit continuity within a predetermined maximum time. Furthermore, to enable the DRA provisioned network to utilize the intact portions of a failed path, the present invention messaging technique provides information to both the origin and destination nodes of a failed path on which spans or links remain intact leading up to the point of failure.

To achieve this end, the failure is first detected by the adjacent custodial nodes bracketing the fault. Each of these custodial nodes adjacent to the failure then would initiate the propagation of a "reuse" message to either the origin node or the destination node. This reuse message has a variable length route information field and an identifier identifying it as a reuse message. As the reuse message is propagated from node to node back to the origin or destination node, each node through which the reuse message passes would append its own unique node identification (ID) to the route information field. Thus, when an origin or destination node receives the reuse message, it can read from the route information field of the reuse message a description of the intact portion of the path. By allowing the restoration logic to take into account the intact portions of the original paths, better restoration decision making use of the intact portions of the path can take place. In one embodiment, such restoration logic is only permitted to apply the intact portions to the restoration of the failed path that originally included the portions. In other words, such intact portions are restricted for use in restoring the failed path. Such restriction considerably simplifies the restoration process and avoids the possibility of unforeseen problems.

The present invention also involves modifying the functionality of each node of the distributed restoration domain, so that, when in receipt of a failed signal (or an AIS signal that suppresses the alarm of downstream equipment), each node of the domain would cause the propagation of a distinct non-alarm signal (or non-AIS signal) that would accomplish the same thing as the original failed signal. Consequently, only the custodial nodes of a failed link, or those nodes that bracket a malfunctioned site, are in receipt of the true alarm or AIS signal.

Since adjacent nodes are connected by links, or spans, the kinds of signals that traverse among the nodes in a network have different formats, depending on the type of connection. In the case of a Digital Service 3 (DS3) facility, each of the nodes of the distributed restoration domain is provisioned with a converter, so that when in receipt ofan AIS signal, the converter would convert the AIS signal into an idle signal (or a modified AIS signal), and propagate the idle signal to nodes downstream thereof.

To achieve this conversion ofan AIS signal, a modification of at least one of the C-bits of the idle signal takes place. Indications of directly adjacent failures such as loss of signal (LOS), loss of frame (LOF) and loss of pointer (LOP) also result in the propagation of a modified idle signal to nodes downstream of where the fault occurred.

At the perimeter of the distributed restoration domain, each of the access/egress nodes that communicatively interconnects the domain to the rest of the network, or other networks, is provisioned such that any incoming modified idle signal is reconverted, or replaced, by a standard AIS signal so that the equipment outside of the distributed restoration domain continue to receive the standard compliant signal.

For those networks that are interconnected by optical fibers where the SONET Synchronous Transport Signal (STS-n) such as the STS 3 standard is used, each node of the distributed restoration domain is provisioned so that, in receipt of an incoming STS-N AIS signal, such AIS signal is replaced by a STS-N Incoming Signal Failure (ISF) signal. In the preferred embodiment, the Z5 bit of the STS-3 signal is changed. This modification serves the same purpose as the C-bit modification to the DS3 signal.

To provide an up-to-date map of the functional spare links of the network, a topology of the network connected by the functional spare links is made available to the custodial nodes that bracket a malfunctioned link as soon as the failure is detected. The custodial node that is designated as the sender or origin node then uses the topology of the spare links to quickly reroute the traffic through the functional spare links.

To ensure that the spare links are functional, prior to the DRA process, special messages, referred to in this invention as keep alive messages, are continuously exchanged on the spare links between adjacent nodes. Each of these keep alive messages has a number of fields which allow it to identify the port of the node from which it is transmitted, the identification of the node, the incoming

IP address and the outgoing IP address of the node, as well as a special field that identifies the keep alive message as coming from a custodial node when there is a detected failure. These keep alive message may be transmitted over the C-bit channels as idle signals. So long as a spare link is operating properly, the keep alive messages that traverse therethrough will contain data that informs the network, possibly by way of the operation support system, of the various pairs of spare ports to which a spare link connects a pair of adjacent nodes.

This information is collected by the network and constantly updated so that at any moment, the network has a view of the entire topology of the network as to what spare links are available. This data can be stored in a database at the operation support system of the network, so that it may be provided to the origin node as soon as a failure is detected.

Additionally, information from so-called keep alive messages utilizing the C-bit as a carrier can be utilized to assigned a "Quality of Service" (QoS)to each link in the network. The QoS can be utilized to assign priorities with respect to how and what data will be re-routed first or if the data will be re-routed at all during a restoration. The QoS can include, among others, quality of performance parameters such as error seconds or severely errored seconds. Once the link has been assigned a value then an algorithm is executed using the data to determine the priority of the data and hence facilitating the restoration process.

One aspect of the present invention can be characterized as providing a special message that is exchanged continuously between adjacent nodes before the occurrence of the failure in order to continually collect data relating to the available spare links of the network.

Another aspect of the present invention can be characterized as providing an improved communication failure detection, isolation and recovery scheme or algorithm.

A further aspect of the present invention can be characterized as providing a method of communicating the description of the intact portions of a failed path to the origin and destination nodes of a failed path. An additional aspect of the present invention includes providing a method of determining the portions of a failed path that remain usable for carrying traffic.

A still further aspect of the present invention can be characterized as identifying the reusable links or spans that connect nodes of a failed path so that these links or spans can be used by an alternate path for restoring the disrupted traffic.

Another aspect of the present invention can be characterized as providing a method of mapping a topology of the spare capacity of a DRA network so that traffic may be routed through the functional spare links when a failure occurs at the network.

Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious f om the description, or may be learned by practice of the invention. The aspects and advantages of the invention will be realized and attained by means o the elements and combinations particularly pointed out in the appended claims.

To achieve these and other advantages, and in accordance with the purpose of the present invention, as embodied and broadly described, the present invention can be characterized according to one aspect as a method for identifying a pair of neighboring nodes in a telecommunications network having at least one distributed restoration sub-network, including constructing a first C-bit keep alive message for a first node in a neighboring node pair connected by a link, embedding the first C-bit keep alive message within the C-bit of a first DS3 signal, determining a quality of service information for the link when looking from the first node to the neighboring node, embedding the quality of service information within the C-bit keep alive message, and transmitting the first DS3 signal from the first node to a second node in the neighboring node pair over the link, wherein the first C-bit keep alive message identifies the first node to the second node and the quality of service information for the link when looking from the first node to the second node.

The present invention can be characterized according to another aspect of the present invention as a telecommunications network including a plurality of nodes interconnected by a plurality of links, and a distributed restoration sub-network, including a first node having a first unique identifier, a second node having a second unique identifier, a link connecting the first node to the second node, and a DS3 signaling channel within the link, wherein the first node and second node are operable to send a DS3 signal having a keep alive message and a quality of service information of the link embedded within a C-bit to one another.

It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the invention, as claimed. BRIEF DESCRIPTION OF THE DRAWINGS

The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate several embodiments of the present invention and, together with the description, serve to explain the principles of the invention. Other objects and advantages will be apparent from a reading of the specification and appended claims in conjunction with the drawings wherein:

FIGURE 1 conceptually illustrates a simplified telecommunications restoration network to provide certain definitions applicable to the present invention;

FIGURE 2 illustrates a restoration subnetwork for illustrating concepts applicable to the present invention;

FIGURE 3 conceptually shows a failure within a restoration subnetwork;

FIGURE 4 illustrates two origins/destination nodes pairs for demonstrating the applicable scope of the present invention;

FIGUREs 5A and 5B illustrate the loose synchronization features of the present invention; FIGURE 6 shows the failure notification message flow applicable to the present invention;

FIGURE 7 illustrates the flow of keep-alive messages according to the present invention;

FIGURE 8 illustrates the flow of path verification messages according to the teachings of the present invention;

FIGURE 9 shows a time diagram applicable to the failure notification and fault isolation process of the present invention;

FIGUREs 10 and 11 illustrate the AIS signal flow within the restoration subnetwork of the present invention;

FIGURE 12 describes more completely the failure notification message flow within the restoration subnetwork according to the present invention; FIGURE 13 illustrates the beginning of an iteration of the restoration process of the present invention;

FIGURE 14 provides a timed diagram applicable to the explore, return, max flow and connect phases of the first iteration of the restoration process of the present invention;

FIGURE 15 provides a timed diagram associated with the explore phase of the process of the present invention;

FIGURE 16 illustrates the possible configuration of multiple origins/destination node pairs from a given origin node;

FIGURE 17 depicts two steps of the explore phase of the first iteration of the restoration process; FIGURE 18 provides a timed diagram applicable to the return phase of the restoration process of the present invention; FIGURE 19 shows steps associated with the return phase of the present process;

FIGUREs 20, 21 and 22 illustrates the link allocation according to the return phase of the present invention;

FIGURE 23 illustrates a typical return message for receipt by the origin node of a restoration subnetwork;

FIGURE 24 provides a timed diagram for depicting the modified map derived from the return messages received at the origin node;

FIGURE 25 illustrates that part of the restoration subnetwork mode within origin node has been allocated to the origin node destination node pair; FIGURE 26 shows the max flow output for the max flow phase of the present process;

FIGURE 27 illustrates an optimal routing applicable to the max flow output of the present invention;

FIGURE 28 provides a timed diagram for showing the sequence of the connect phase for the first iteration of the process of the present invention; FIGURE 29 illustrates the connect messages for providing the alternate path routes between an origin node and destination node of a restoration subnetwork;

FIGUREs 30 and 31 show how the present invention deals with hybrid restoration subnetworks;

FIGUREs 32 and 33 illustrate the explore phase and return phase, respectively, applicable to hybrid networks;

FIGURE 34 shows the time diagram including an extra iteration for processing hybrid networks according to the teachings of the present invention;

FIGUREs 35 and 36 illustrate a lower quality spare according to the teachings of the present invention; FIGURE 37 illustrate the use of a "I am custodial node" flag of the present invention;

FIGUREs 38 through 42 describe the restricted re-use features of the present invention;

FIGURE 43 describes the path inhibit feature of the present invention;

FIGURE 44 further describes the path inhibit feature of the present invention;

FIGURE 45 is a path of a telecommunications network for illustrating the instant invention; FIGURE 46 illustrates the messages that are sent from the custodial nodes of the failed path of FIGURE 1;

FIGURE 47 is another view of the failed path of FIGURE 45 in which messages are sent from the intermediate nodes to their respective downstream nodes; -,

FIGURE 48 is the failed path of FIGURE 45 showing a message reaching the destination node, the interconnections between the origin and destination nodes, and the use of a spare link for bypassing the fault; and FIGURE 49 is an illustration of the reuse message of the present invention.

FIGURE 50 is an illustration of a telecommunications network of the instant invention;

FIGURE 51 is a block diagram illustrating two adjacent cross-connect switches and the physical interconnection therebetween; and FIGURE 52 is an illustration of the structure ofan exemplar keep alive message of the present invention.

FIGURE 53 shows a plurality of nodes of a distributed restoration domain through which an alarm signal generated as a result of a malfunction at a link interconnecting two of the nodes is shown to be propagated to downstream nodes; FIGURE 54 illustrates the same nodes as shown in the FIGURE 53 DS3 environment, but in this instance those nodes of the distributed restoration domain each are provisioned to convert an incoming alarm signal into a non-alarm signal, and the access/egress nodes of the distributed restoration domain are further provisioned to reconvert any received modified alarm signal back into an alarm signal; FIGURE 55 shows the frame structure of a DS3 signal for illustrating the conversion of an alarm signal into a non-alarm signal in a DS3 format;

FIGURE 56 is an illustration that is similar to the FIGURE 54 embodiment, except that the FIGURE 56 embodiment illustrates a SONET network;

FIGURE 57 shows the format of a STS-3 frame for explaining the conversion ofan alarm signal into a non-alarm signal in a SONET network;

FIGURE 58 is a simplified block diagram illustrating the flow of a signal into a node of a distributed restoration domain;

FIGURE 59 is a block representation of a node in the distributed restoration domain of the instant invention provisioned to convert an alarm signal into a non-alarm signal, and in the case ofan access/egress node, to reconvert a non-alarm signal back into an alarm signal;

FIGURE 60 is a graph demonstrating the statuses of the different signals into and out of the nodes of a distributed restoration domain; and

FIGURE 61 is a block representation depicting a communications network with multiple interconnect nodes connected via links and ports of the present invention.

DETAILED DESCRIPTION OF THE INVENTION

FIGURE 1 shows telecommunications network portion 10, that includes node 12 that may communicate with node 14 and node 16, for example. Connecting between node 12 and 14 may be a set of links such as links 18 through 26, as well as for example, links 28 through 30 between node 12 and node 16. Node 14 and node 16 may also communicate between one another through links 32 through 36, for example, which collectively may be thought of as a span 38. The following description uses certain terms to describe the concepts of the present invention. The term 1633SX is a cross-connect switch and is here called a "node." Between nodes are links, which may be a DS3, and STS-1, which is essentially the same thing as a DS3, but which conforms to a different standard. A link could be an STS-3, which is three STS-ls multiplexed together to form a single signal. A link may also be a STS-12, which is twelve STS-ls multiplexed together, or a link could be an STS-12C, which is twelve STS-12s, which are actually locked together to form one large channel. A link, however, actually is one unit of capacity for the purposes of the present invention. Thus, for purposes of the following description, a link is a unit of capacity connecting between one node and another. A span is to be understood as all of the links between two adjacent nodes. Adjacent nodes or neighbor nodes are connected by a bundle, which itself is made up of links.

For purposes of the present description, links may be classified as working, spare, fail, or recovered. A working link is a link that currently carries traffic. Spare links are operable links that are not currently being used. A spare link may be used whenever the network desires to use the link. A failed link is a link that was working, but has failed. A recovered link is a link that, as will be described more completely below, has been recovered.

FIGURE 2 illustrates the conceptual example of restoration subnetwork 40 that may include origin node 42 that through tandem nodes 44 and 46 connects to destination node 48. In restoration subnetwork 40, a path such as paths 50, 52, 54, and 56 includes connections to nodes 42 through 48, for example, as well as links between these nodes. As restoration subnetwork 40 depicts, each of the paths enters restoration subnetwork 40 from outside restoration subnetwork 40 at origin node 42. With the present embodiment, each of nodes 42 through 48 includes an associated node identifier. Origin node 42 possesses a lower node identifier value, while destination node 48 possesses a higher node identifier value. In the restoration process of the present invention, the nodes compare node identification numbers. The present invention establishes restoration subnetwork 40 that may be part of an entire telecommunications network 10. Within restoration subnetwork 40, there may be numerous paths 50. A path 50 includes a number of links 18 strung together and crossconnected through the nodes 44. The path 50 does not start within restoration subnetwork 40, but may start at a customer premise or someplace else. In fact, a path 50 may originate outside a given telecommunications network 10. The point at which the path 50 enters the restoration subnetwork 40, however, is origin node 42. The point on origin node 42 at which path 50 comes into restoration subnetwork 40 is access/egress port 58.

In a restoration subnetwork, the failure may occur between two tandem nodes. The two tandem nodes on each side of the failure are designated as "custodial" nodes. If a single failure occurs in the network, there can be two custodial nodes. In the network, therefore, there can be many origin/destination nodes. There will be two origin nodes and two destination nodes. An origin node together with an associated destination node may be deemed an origin/destination pair. One failure may cause many origin/destination pairs.

FIGURE 3 illustrates the concept of custodial nodes applicable to the present invention. Referring again to restoration subnetwork 40, custodial nodes 62 and 64 are the tandem nodes positioned on each side of failed span 66. Custodial nodes 62 and 64 have bound the failed link and communicate this failure, as will be described below. FIGURE 4 illustrates the aspect of the present invention for handling more than one origin-destination node pair in the event of a span failure. Referring to FIGURE 4, restoration subnetwork 40 may include, for example, origin node 42 that connects through custodial nodes 62 and 64 to destination node 48. Within the same restoration subnetwork, there may be more than one origin node, such as origin node 72. In fact, origin node 72 may connect through custodial node 62 and custodial node 64 to destination node 74. As in FIGURE 3, FIGURE 4 shows failure 66 that establishes custodial nodes 62 and 64.

The present invention has application for each origin/destination pair in a given restoration subnetwork. The following discussion, however, describes the operation of the present invention for one origin/destination pair, obtaining an understanding of how the present invention handles a single origin/destination pair makes clear how the algorithm may be extended in the event of several origin/destination pairs occurring at the same time. An important consideration for the present invention, however, is that a single cut may produce numerous origin/destination pairs.

FIGUREs 5A and 5B illustrate the concept of loose synchronization according to the present invention. "Loose synchronization" allows operation of the present method and system as though all steps were synchronized according to a centralized clock. Known restoration algorithms suffer from race conditions during restoration that make operation of the restoration process unpredictable. The restoration configuration that results in a given network, because of race conditions, depends on which messages arrive first. The present invention eliminates race conditions and provides a reliable result for each given failure. This provides the ability to predict how the restored network will be configured, resulting in a much simpler restoration process.

Referring to FIGURE 5A, restoration subnetwork 40 includes origin node 42, that connects to tandem nodes 44 and 46. Data may flow from origin node 42 to tandem node 46, along data path 76, for example. Origin node 42 may connect to tandem node 44 via path 78. However, path 80 may directly connect origin node 42 with destination node 48. Path 82 connects between tandem node 44 and tandem node 46. Moreover, path 84 connects between tandem node 46 and destination node 48. As FIGURE 5A depicts, data may flow along path 76 from origin node 42 to tandem node 46, and from destination node 48 to origin node 42. Moreover, data may be communicated between tandem node 44 and tandem node 46. Destination node 48 may direct data to origin node 42 along data path 80, as well as to tandem node 46 using path 84. These data flows will all take place in a single step. At the end of a step, each of the nodes in restoration subnetwork 40 sends a "step complete" message to its neighboring node. Continuing with the example of FIGURE 5A, in FIGURE 5B there are numerous step complete messages that occur within restoration subnetwork 40. In particular, step complete message exchanges occur between origin node 42 and tandem node 44 on data path 78, between origin node 42 and tandem node 46 on data path 76, and between origin node 42 and destination node 48 on data path 80. Moreover, tandem node 46 exchanges "step complete" messages with tandem node 44 on data path 82, and between tandem node 46 and destination node 48 on data path 84.

In the following discussion, the term "hop count" is part of the message that travels from one node to its neighbor. Each time a message flows from one node to its neighbor, a "hop" occurs. Therefore, the hop count determines how many hops the message has taken within the restoration subnetwork.

The restoration algorithm of the present invention may be partitioned into steps. Loose synchronization assures that in each step a node processes the message it receives from its neighbors in that step. Loose synchronization also makes the node send a step complete message to every neighbor. If a node has nothing to do in a given step, all it does is send a step complete message. When a node receives a step complete message from all of its neighbors, it increments a step counter associated with the node and goes to the next step.

Once a node receives step complete messages from every neighbor, it goes to the next step in the restoration process. In looking at the messages that may go over a link, it is possible to see a number of messages going over the link. The last message, however, will be a step complete message. Thus, during the step, numerous data messages are exchanged between nodes. At the end of the step, all the nodes send step complete messages to their neighbors to indicate that all of the appropriate data messages have been sent and it is appropriate to go to the next step. As a result of the continual data, step complete, data, step complete, message traffic, a basic synchronization occurs. In practice, although the operation is not as synchronized as it may appear in the associated FIGUREs, synchronization occurs. During the operation of the present invention, messages travel through the restoration subnetwork at different times. However, loose synchronization prevents data messages from flowing through the restoration subnetwork until all step complete messages have been received at the nodes. It is possible for one node to be at step 3, while another node is at step 4. In fact, at some "places within the restoration subnetwork, there may be even further step differences between nodes. This helps minimize the effects of slower nodes on the steps occurring within the restoration subnetwork.

The steps in the process of the present invention may be thought of most easily by considering them to be numbered. The process, therefore, starts at step 1 and proceeds to step 2. There are predetermined activities that occur at each step and each node possesses its own step counter. However, there is no master clock that controls the entire restoration subnetwork. In other words, the network restoration process of the present invention may be considered as a distributive restoration process. With this configuration, no node is any different from any other node. They all perform the same process independently, but in loose synchronization. FIGURE 6 shows the typical form of a failure notification message through restoration subnetwork 40. If, for example, origin node 42 desires to start a restoration event, it first sends failure notification messages to tandem node 44 via data path 78, to tandem node 46 via data path 76, and destination node 48 via data path 80. As FIGURE 6 further shows, tandem node 44 sends failure notification message to tandem node 46 on path 82, as does destination node 48 to tandem node 46 on path, 84.

The process of the present invention, therefore, begins with a failure notification message. The failure notification message is broadcast throughout the restoration subnetwork to begin the restoration process from one node to all other nodes, once a node receives a failure message, it sends the failure notification message to its neighboring node, which further sends the message to its neighboring nodes. Eventually the failure notification message reaches every node in the restoration subnetwork. Note that if there are multiple failures in a network, it is possible to have multiple failure notification messages flooding throughout the restoration subnetwork simultaneously.

The first failure notification message initiates the restoration algorithm of the present invention. Moreover, broadcasting the failure notification message is asynchronous in the sense that as soon as the node receives the failure notification message, it broadcasts the message to its neighbors without regard to any timing signals. It is the failure notification message that begins the loose synchronization process to begin the restoration process of the present invention at each node within the restoration subnetwork. Once a node begins the restoration process, a series of events occurs. Note, however, that before the restoration process of the present invention occurs, numerous events are already occurring in the restoration subnetwork. One such event is the transmission and receipt of keep alive messages that neighboring nodes exchange between themselves.

FIGURE 7 illustrates the communication of keep-alive messages that the restoration process of the present invention communicates on spare links, for the purpose of identifying neighboring nodes. Referring to FIGURE 7, configuration 90 shows the connection via spare link 92 between node 94 and node 96. Suppose, for example, that node 94 has the numerical designation 111", and port designation 11103". Suppose further that node 96 has the numerical designation 3 and the port designation 5. On spare link 92, node 94 sends keep-alive message 98 to node 96, identifying its node number "11" and port number "103". Also, from node 96, keep-alive message 100 flows to node 94, identifying the keep-alive message as coming from the node having the numerical value "3", and its port having the numerical value "5". The present invention employs keep-alive signaling using C-Bit of the DS3 formatted messages in restoration subnetwork 40, the available spare links carry DS3 signals, wherein the C-bits convey special keep-alive messages. In particular, each keep-alive message contains the node identifier and port number that is sending the message, the WAN address of the node, and an "I am custodial node" indicator to be used for assessing spare quality.

An important aspect of the present invention relates to signaling channels which occurs when cross-connect nodes communicate with one another. There are two kinds of communications the cross-connects can perform. One is called in-band, another is out-of-band. With in-band communication, a signal travels over the same physical piece of media as the working traffic. The communication travels over the same physical media as the path or the same physical media as the link. With out-of-band signals, there is freedom to deliver the signals between cross-connects in any way possible. Out-of-band signals generally require a much higher data rate.

In FIGURE 7, for example, in-band messages are piggybacked on links, out-of-band message traffic may flow along any other possible path between two nodes. With the present invention, certain messages must flow in-band. These include the keep-alive message, the path verification message, and the signal fail message. There are some signaling channels available to the restoration process of the present invention, depending on the type of link involved. This includes SONET links and asynchronous links, such as DS3 links.

A distinguishing feature between SONET links and DS3 links is that each employs a different framing standard for which unique and applicable equipment must conform. It is not physically possible to have the same port serve as a SONET port and as a DS3 port at the same time. In SONET signal channeling, there is a feature called tandem path overhead, which is a signaling channel that is part of the signal that is multiplexed together. It is possible to separate this signal portion from the SONET signaling channel. Because of the tandem path overhead, sometimes called the Z5 byte, there is the ability within the SONET channel to send messages.

On DS3 links, there are two possible signaling channels. There is the C-bit and the X-bit. The C-bit channel cannot be used on working paths, but can only be used on spare or recovered links. This is because the DS3 standard provides the option using the C-bit or not using the C-bit. If the C- bit format signal is used, then it is possible to use the C-bit for signaling. However, in this instance, working traffic does not use that format. Accordingly, the C-bit is not available for signaling on the working channels. It can be used only on spare links and on recovered links.

FIGURE 8 illustrates in restoration subnetwork 40 the flow of path verification messages from origin node 42 through tandem nodes 44 and 46 to destination node 48. Path verification message 102 flows from origin node 42 through tandem nodes 44 and 46 to destination node 48. In particular, suppose origin node 42 has the label 18, and that working path 52 enters port 58. Path verification message 102, therefore, contains the labels 18 and 53, and carries this information through tandem nodes 44 and 46 to destination node 48. Destination node 48 includes the label 15 and egress port 106 having the label 29. Path verification message 104 flows through tandem node 46 and 44 to origin node 42 for the purpose of identifying destination node 48 as the destination node for working path 52. A path verification message is embedded in a DS3 signal using the X-bits which are normally used for very low speed single-bit alarm signaling. In the present invention, the X-bit state is overridden with short bursts of data to communicate signal identity to receptive equipment downstream. The bursts are of such short duration that other equipment relying upon traditional use of the X-bit for alarm signaling will not be disturbed. The present invention also provides for confining path verification signals within a network.

In a DRA controlled network, path verification messages are imbedded in traffic-bearing signals entering the network and removed from signals leaving the network. Inside of the network, propagation of such signals is bounded based upon the DRA-enablement status of each port. The path verification messages identify the originating node and the destination node. The path verification messages occur on working links that are actually carrying traffic. The path verification message originates at origin node 42 and the restoration subnetwork and passes through tandem nodes until the traffic reaches destination node 48. Tandem nodes 44 and 46 between the origin node 42 and destination node 48, for example, can read the path verification message but they cannot modify it. At destination node 48, the path verification message is stripped from the working traffic to prevent its being transmitted from the restoration subnetwork.

The present invention uses the X-bit to carry path verification message 104. one signal format that the present invention may use is the DS3 signal format. While it is possible to easily provide a path verification message on SONET traffic, the DS3 traffic standard does not readily permit using path verification message 104. The present invention overcomes this limitation by adding to the DS3 signal, without interrupting the traffic on this signal and without causing alarms throughout the network, path verification message 104 on the DS3 frame X-bit.

The DS3 standard specifies that the signal is provided in frames. Each frame has a special bit in it called the X-bit. In fact, there are two X-bits, X-l and X-2. The original purpose of the X-bit, however, was not to carry path verification message 104. The present invention provides in the X-bit the path verification message. This avoids alarms and equipment problems that would occur if path verification message 104 were placed elsewhere. An important aspect of using the X-bit for path- verification message 104 with the present embodiment relates to the format of the signal. The present embodiment sends path verification message 104 at a very low data rate, for example, on the order of five bits per second. By sending path verification message 104 on the X-bit very slowly, the possibility of causing an alarm in the network is significantly reduced. Path verification message 104 is sent at a short burst, followed by a long waiting period, followed by a short burst, followed by a long waiting period, etc. This method of "sneaking" path verification message 104 past the alarms permits using path verification message 104 in the DS3 architecture systems.

FIGURE 9 shows conceptually a timeline for the restoration process that the present invention performs. With time moving downward, time region 108 depicts the network status prior to a failure happening at point 110. At the point that a failure happens, the failure notification and fault isolation events occur in time span 112. Upon completion of this step, the first generation of the present process occurs, as indicated by space 114. This includes explore phase 116 having, for example three steps 118, 120 and 122. Return phase 124 occurs next and may include at least two steps 126 and 128. These steps are discussed more completely below. Once a failure occurs, the process of the present invention includes failure notification and fault isolation phase 112. Failure notification starts the process by sending failure notification messages throughout the restoration subnetwork. Fault isolation entails determining which nodes are the custodial nodes. One reason that it is important to know the custodial nodes is that there are spares on the same span as the failed span. The present invention avoids using those spares, because they are also highly likely to fail. Fault isolation, therefore, provides a way to identify which nodes are the custodial nodes and identifies the location of the fault along the path.

FIGURE 10 illustrates the flow of AIS signals 130 through restoration subnetwork 40. In the event of failure 66 between custodial nodes 62 and 64, the AIS message 130 travels through custodial node 62 to origin node 42 and out restoration subnetwork 40. Also, AIS message 130 travels through custodial node 64 and tandem node 46, to destination node 48 before leaving restoration subnetwork 40. This is the normal way of communicating AIS messages 130. Thus, normally every link on a failed path sees the same AIS signal.

FIGURE 11, on the other hand, illustrates the conversion of AIS signal 130 to "signal fail,, signals 132 and 134. SF message 132 goes to origin node 42, at which point it is reconverted to AIS message 132. Next, signal 134 passes through tandem node 46 en route to destination node 48,-which reconverts SF message 134 to AIS message 130.

FIGUREs 10 and 11, therefore, illustrate how the DS3 standard specifies operations within the restoration subnetwork. For a DS3 path including origin node 42 and destination node 48, with one or more tandem nodes 44, 46. Custodial nodes 62 and 64 are on each side of the link failure 66. AIS signal 130 is a DS3 standard signal that indicates that there is an alarm downstream. Moreover, AIS signal 130 could actually be several different signals. AIS signal 130 propagates downstream so that every node sees exactly the same signal.

With AIS signal 130, there is no way to determine which is a custodial node 62, 64 and which is the tandem node 44, 46. This is because the incoming signal looks the same to each receiving node. The present embodiment takes this into consideration by converting AIS signal 130 to a signal fail or SF signal 132. When tandem node 46 sees SF signal 134, it propagates it through until it reaches destination node 48 which converts SF signal 134 back to AIS signal 130.

Another signal that may propagate through the restoration subnetwork 40 is the ISF signal. The ISF signal is for a signal that comes into the restoration subnetwork and stands for incoming signal fail. An ISF signal occurs if a bad signal comes into the network, if it comes in as an AIS signal, there is the need to distinguish that, as well. In the SONET standard there is already an ISF signal. The present invention adds the SF signal, as previously mentioned. In the DS3 standard, the SF signal already exists. The present invention adds the ISF signal to the DS3 standard. Consequently, for operation of the present invention in the DS3 standard environment, there is the addition of the ISF signal. For operation in the SONET standard environment, the present invention adds the SF signal. Therefore, for each of the standards, the present invention adds a new signal.

To distinguish whether an incoming non-traffic signal received by a node has been asserted due to an alarm within a DRA-controlled network, a modified DS3 idle signal is propagated downstream in place of the usual Alarm Indication Signal (AIS). This alarm-produced idle signal differs from a normal idle signal by an embedded messaging in the C-bit maintenance channel to convey the presence of a failure within the realm of a particular network. The replacement of AIS with idle is done to aid fault isolation by squelching downstream alarms. Upon leaving the network, such signals may be converted back into AIS signals to maintain operational compatibility with equipments outside the network. A comparable technique is performed in a SONET network, where STS-N AIS signals are replaced with ISF signal and the ZS byte conveys the alarm information.

Another aspect of the present invention is the ability to manage unidirectional failures. In a distributed restoration environment, failures that occur along one direction of a bi-directional link are handled by first verifying that the alarm signal persists for a period of time and then propagating an idle signal back along the remaining working direction. This alarm produced idle signal differs from a normal idle signal by embedded messaging in the C-bit maintenance channel to convey the presence of a far end receive failure. In this manner, custodial nodes are promptly identified and restorative switching is simplified by treating unidirectional failures as if they were bi-directional failures.

FIGURE 12 illustrates the broadcast of failure notification messages from custodial nodes 62 and 64. As FIGURE 12 depicts, custodial node 62 sends a failure notification to origin node 42, as well as to tandem node 136. Tandem node 136 further broadcasts the failure notification message to tandem nodes 138 and 140. In addition, custodial node 64 transmits a failure notification message to tandem node 46, which further transmits the failure notification message to destination node 48. Also, custodial node 64 broadcasts the failure notification message to tandem node 140.

FIGURE 13 illustrates the time diagram for the first iteration following fault isolation. In particular, FIGURE 13 shows the time diagram for explore phase 116 and return phase 124 of iteration 1. FIGURE 14 further illustrates the time diagram for the completion of iteration 1 and a portion of iteration 2. As FIGURE 14 indicates, iteration 1 includes explore phase 116, return phase 124, max flow phase 142 and connect phase 144. Max flow phase 142 includes a single step 146. Note that connect phase 144 of iteration 2 shown by region 148 includes six steps, 150 through 160, and occurs simultaneously with explore phase 162 of iteration 2. Note further that return phase 164 of iteration 2 also includes six steps 166 through 176.

Each iteration involves explore, return, maxflow, and connect phases. The restored traffic addressed by connect message and the remaining unrestored traffic conveyed by the explore message are disjoint sets. Hence, there is no conflict in concurrently propagating or combining these messaging steps in a synchronous DRA process. In conjunction with failure queuing, this practice leads to a restoration process that is both reliably coordinated and expeditious.

The iterations become longer in duration and include more steps in subsequent iterations. This is because with subsequent iterations, alternate paths are sought. A path has a certain length in terms of hops. A path may be three hops or four hops, for example. In the first iteration, for example, a hop count may be set at three. This, means that alternate paths that are less than or equal to three hops are sought. The next iteration may seek alternate paths that are less than or equal to six hops. Setting a hop count limit per iteration increases the efficiency of the process of the present invention. With the system of the present invention, the number of iterations and the number of hop counts for each iteration is configurable. However, these may also be preset, depending on the degree of flexibility that a given implementation requires. Realize, however, that with increased configurability, increased complexity results. This increased complexity may, in some instances, generate the possibility for inappropriate or problematic configurations.

FIGURE 15, for promoting the more detailed discussion of the explore phase, shows explore phase 116, which is the initial part of the first iteration 114. FIGURE 16 shows restoration network portion 170 to express the idea that a single origin node 42 may have more than one destination node. In particular, destination node 180 may be a destination node for origin node 42 through custodial nodes 62 and 66. Also, as before, destination node 48 is a destination node for origin node 42. This occurs because two working paths, 182 and 184, flow through restoration subnetwork portion 170, both beginning at origin node 42. During the explore phase, messages begin at the origin nodes and move outward through the restoration subnetwork. Each explore message is stored and forwarded in a loosely synchronized manner. Accordingly, if a node receives the message in step 1, it forwards it in step 2. The neighboring node that receives the explore message in step 1 transmits the explore message to its neighboring node in step 2. Because the present invention employs loose synchronization it does not matter how fast the message is transmitted from one neighbor to another, it will be sent at the next step irrespectively. If the explore phase is three steps long, it may flood out three hops and no more. The following discussion pertains to a single origin-destination pair, but there may be other origin/destination pairs performing the similar or identical functions at the same time within restoration subnetwork 40. If two nodes send the explore message to a neighboring node, only the first message received by the neighboring node is transmitted by the neighboring node. The message that is second received by the neighboring node is recognized, but not forwarded. Accordingly, the first node to reach a neighboring node with an explore message is generally the closest node to the neighboring node. When an explore message reaches the destination node, it stops. This step determines the amount of spare capacity existing in the restoration subnetwork between the origin node and the destination node.

Because of loose synchronization, the first message that reaches origin node 42 and destination node 48 will be the shortest path. There are no race conditions within the present invention's operation. In the explore message, the distance between the origin node and destination node is included. This distance, measured in hops, is always equal to or less than the number of steps allowed for the given explore phase. For example, if a destination node is five hops from the origin node by the shortest path, the explore phase with a three hop count limit will never generate a return message. On the other hand, an explore phase with a six hop count limit will return the five hop count information in the return message.

In the explore message there is an identification of the origin-destination pair to identify which node sent the explore message and the destination node that is to receive the explore message. There is also a request for capacity. The message may say, for example, that there is the need for thirteen DS3s, because thirteen DS3s failed. In practice, there may be not just DS3s, but also STS-ls, STS-12C's, etc. The point being, however, that a certain amount of capacity is requested. At each node that the explore message passes through, the request for capacity is noted. The explore phase is over once the predetermined number of steps have been completed. Thus, for example, if the explore phase is to last three steps, at step 4, the explore phase is over. This provides a well-defined end for the explore phase.

FIGURE 17 illustrates restoration subnetwork 40 for a single-origin destination pair, including origin node 42 and destination node 48. In restoration subnetwork 40, origin node 42, at the beginning of the explore phase, takes step 1 to send an explore message to tandem node 44, tandem node 46 and tandem node 186. At step 2, tandem node 46 sends an explore message to tandem node 188 and to destination node 48. At step 2, tandem node 44 sends an explore message to tandem node 46, tandem node 46 sends an explore message to tandem node 188, and to destination node 48, and tandem node 186 sends explore messages to tandem node 46 and to destination node 48. Note that explore messages at step 2 from tandem node 44 to tandem node 46 and from tandem node 186 to tandem node 46 are not forwarded by tandem node 46. FIGURE 18 illustrates the time diagram for the next phase in the restoration process of the present invention, the return phase 24, which during the first iteration, includes three steps, 126, 128 and 129.

FIGURE 19 illustrates the return phase of the present invention, during the first iteration. Beginning at destination node 48, at step 4, return message flows on path 192 to tandem node 46, and on path 190 to tandem node 186. At step 5, the return message flows on path 76 to origin node 42. Also, from tandem node 186, a return message flows to origin node 42.

During the return phase, a return message flows over the same path traversed by its corresponding explore phase, but in the opposite direction. Messages come from the destination node and flow to the origin node. In addition, the return phase messages are loosely synchronized as previously described. The return phase messages contain information relating to the number of spare links available for connecting the origin node to the destination node.

In the return phase, information relating to the available capacity goes to the origin node. Beginning at destination node 48, and continuing through each tandem node 44, 46, 186 en route to origin node 42, the return message becomes increasingly longer. The return message, therefore, contains information on how much capacity is available on each span en route to the origin node. The result of the return message received is the ability to establish at the origin node a map of the restoration network showing where the spare capacity is that is useable for the restoration.

FIGURE 20 illustrates tandem node 44, that connects to tandem node 46 through span 38. Note that span 38 includes six links 32, 34, 36, 196, 198 and 200. FIGUREs 21 and 22 illustrate the allocation of links between the tandem nodes 44, 46 according to the preferred embodiment of the present invention. Referring first to FIGURE 21, suppose that in a previous explore phase, span 38 between tandem nodes 44 and 46 carries the first explore message (5,3) declaring the need for four links for node 46, such as scenario 202 depicts. Scenario 204 shows further a message (11,2) requesting eight link flows from tandem node 44, port 2.

FIGURE 22 illustrates how the present embodiment allocates the six links of span 38. In particular, in response to the explore messages from scenarios 202 and 204 of FIGURE 21, each of tandem nodes 44 and 46 knows to allocate three links for each origin destination pair. Thus, between tandem nodes 44 and 46, three links, for example links 32, 34 and 36 are allocated to the (5,3) origin destination pair. Links 196, 198 and 200, for example, may be allocated to the origin/destination pair (11,2).

FIGURE 23 illustrates the results of the return phase of the present invention. Restoration subnetwork 40 includes origin node 42, tandem nodes 208, 210 and 212, as well as tandem node 44, for example. As FIGURE 23 depicts, return messages carry back with them a map of the route they followed and how much capacity they were allocated on each span. Origin node 42 collects all the return messages. Thus, in this example, between origin node 42 and tandem node 44, four links were allocated between origin node 42 and node 208. Tandem node 208 was allocated ten links to tandem node 210. Tandem node 210 is allocated three links, with tandem node 17. And tandem node 17 is allocated seven links with tandem node 44.

The next phase in the first iteration of the process of the present invention is the maxflow phase. The maxflow is a one-step phase and, as FIGURE 24 depicts, for example, is the seventh step of the first iteration. All of the work in the maxflow phase for the present embodiment occurs at origin node 42. At the start of the maxflow phase, each origin node has a model of part of the network. This is the part that has been allocated to the respective origin/destination pair by the tandem nodes. FIGURE 25 illustrates that within origin node 42 is restoration subnetwork model 214, which shows what part of restoration subnetwork 40 has been allocated to the origin node 42-destination node 48 pair. In particular, model 214 shows that eight links have been allocated between origin node 42 and tandem node 46, and that eleven links have been allocated between tandem node 46 and destination node 48. Model 214 further shows that a possible three links may be allocated between tandem node 46 and tandem node 186.

As FIGURE 26 depicts, therefore, in the maxflow phase 142 of the present embodiment, origin node 42 calculates alternate paths through restoration subnetwork 40. This is done using a maxflow algorithm. The maxflow output of FIGURE 26, therefore, is a flow matrix indicating the desired flow of traffic between origin node 42 and destination node 48. Note that the maxflow output uses neither tandem node 44 nor tandem node 188.

FIGURE 27 illustrates a breadth-first search that maxflow phase 142 uses to find routes through the maxflow phase output. In the example in FIGURE 27, the first route allocates two units, first from origin node 42, then to tandem node 186, then to tandem node 46, and finally to destination node 48. A second route allocates three units, first from origin node 42 to tandem node 186, and finally to destination node 48. A third route allocates eight units, first from origin node 42 to tandem node 46. From tandem node 46, these eight units go to destination node 48.

The last phase in the first iteration in the process of the present embodiment includes connect phase 144. For the example herein described, connect phase includes steps 8 through 13 of the first iteration, here having reference numerals 150, 152, 154, 156, 220 and 222, respectively. The connect phase is loosely synchronized, as previously described, such that each connect message moves one hop in one step. Connect phase 144 overlaps explore Phase 162 of each subsequent next iteration, except in the instance of the last iteration. Connect phase 144 distributes information about what connections need to be made from, for example, origin node 42 through tandem nodes 46 and 186, to reach destination node 48. In connect phase 144, messages flow along the same routes as identified during maxflow phase 142. Thus, as FIGURE 29 suggests, a first message, Ml, flows from origin node 42 through tandem node 186, through tandem node 46 and finally to destination node 48, indicating the connection for two units. Similarly, a second message, M2, flows from origin node 42 through tandem node 186 and then directly to destination node 48, for connecting a three-unit flow path. Finally, a third connect message, M3, emanates from origin node 42 through tandem node 46, and then the destination node 48 for allocating eight units. Connect phase 144 is synchronized so that each step in a message travels one hop.

For implementing the process of the present invention in an existing or operational network, numerous extensions are required. These extensions take into consideration the existence of hybrid networks, wherein some nodes have both SONET and DS3 connections. Moreover, the present invention provides different priorities for working paths and different qualities for spare links. Fault isolation presents a particular challenge in operating or existing environments, that the present invention addresses. Restricted reuse and spare links connected into paths are additional features that the present invention provides. Inhibit functions such as path-inhibit and node-inhibit are additional features to the present invention. The present invention also provides features that interface with existing restoration processes and systems, such as coordination with an existing restoration algorithm and process or similar system. To ensure the proper operation of the present invention, the present embodiment provides an exerciser function for exercising or simulating a restoration process, without making the actual connections for subnetwork restoration. Other features of the present implementation further include a drop-dead timer, and an emergency shutdown feature to control or limit restoration subnetwork malfunctions. Additionally, the present invention handles real life situations such as glass-throughs and staggered cuts that exist in communications networks. Still further features of the present embodiment include a hold-off trigger, as well as mechanisms for hop count and software revision checking, and a step timer to ensure proper operation.

FIGUREs 30 through 33 illustrate how the present embodiment addresses the hybrid networks. A hybrid network is a combination of asynchronous and SONET links. Restrictions in the way that the present invention handles hybrid networks include that all working paths must either be SONET paths with other than DS3 loading, or DS3 over asynchronous and SONET working paths with DS3 access/egress ports. Otherwise, sending path verification messages within the restoration subnetwork 40, for example, may not be practical. Referring to FIGUREs 30 and 31 , restoration subnetwork 40 may include SONET origin A/E port 42, that connects through SONET tandem port 44, through sonnet tandem port 46 and finally to sonnet destination A/E port 48. In FIGURE 31 , origin A/E port 42 is a DS3 port, with tandem port 44 being a sonnet node, and tandem port 46 being a DS3 port, for example. Port 106 of destination node 48 is a DS3 port. In a hybrid network, during the explore phase, origin node 42 requests different types of capacity. In the return phase, tandem nodes 44, 46 allocate different types of capacity. An important aspect of connect phase 144 is properly communicating in the connect message the type of traffic that needs to be connected. This includes, as mentioned before, routing DS3s, STS- ls, OC-3s, and OC-12Cs, for example. There is the need to keep track of all of the implementation details for the different types of traffic. For this purpose, the present invention provides different priorities of working paths and different qualities of spare links. With the present embodiment of the invention, working traffic is prioritized between high priority and low priority working traffic.

SONET traffic includes other rules to address as well. For instance, a SONET path may include an OC-3 port, which is basically three STS-1 ports, with an STS-1 representing the SONET equivalent of a DS3 port. Thus, an OC-3 node can carry the same traffic as can three STS-1. An OC- 3 node can also carry the same traffic as three DS3s or any combination of three STS-1 and DS3 nodes. In addition, an OC-3 node may carry the same traffic as an STS-3. So, an OC-3 port can carry the same traffic as three DS3, three STS-1, or one OC-3. Then, an OC-12 may carry an OC-12C. It may also carry the same traffic as up to four OC-3 ports, up to 12 STS-1 ports, or up to twelve DS3 ports. With all of the possible combinations, it is important to make sure that the large capacity channels flow through the greatest capacity at first.

An important aspect of the present invention, therefore, is its ability to service hybrid networks. A hybrid network is a network that includes both SONET and asynchronous links, such as DS3 links. The present invention provides restoration of restoration subnetwork 40 that may include both types of links. The SONET standard provides that SONET traffic is backward compatible to DS3 traffic. Thus, a SONET link may include a DS3 signal inside it. A restoration subnetwork that includes both SONET and DS3 can flow DS3-signals, provided that both the origin A/E port 42 and the destination A E port 48 are DS3 ports. If this were not the case, there would be no way to send path verification messages 104 within restoration subnetwork 40.

As with pure networks, with hybrid networks, explore messages request capacity for network restoration. These messages specify what kind of capacity that is necessary. It is important to determine whether DS3 capacity or SONET capacity is needed. Moreover, because there are different types of SONET links, there is the need to identify the different types of format of SONET that are needed. In the return phase, tandem nodes allocate capacity to origin-destination pairs. Accordingly, they must be aware of the type of spares that are available in the span. There are DS3 spares and SONET spares. Capacity may be allocated knowing which type of spares are available. There is the need, therefore, in performing the explore and return phases, to add extensions that allow for different kinds of capacity. The explore message of the present invention, therefore, contains a request for capacity and decides how many DS3s and how many SONET links are necessary. There could be the need for an STS-1, an STS-3C, or an STS-12C, for example. Moreover, in the return phase it is necessary to include in the return message the information that there is more than one kind of capacity in the network. When traffic routes through the network it must be aware of these rules. For instance, a DS3 failed working link can be carried by a SONET link, but not vice versa. In other words, a DS3 cannot carry a SONET failed working path.

FIGUREs 32 and 33 illustrate this feature. For example, referring to FIGURE 32, origin node 42 may generate explore message to tandem node 44 requesting five DS3s, three STS-ls, two STS- 3(c)s, and one STS-12(c)s. As FIGURE 33 depicts, from the return phase, origin node 42 receives return message from tandem node 44, informing origin node 42 that it received five DS3s, one STS-1, one STS-3(c), and no STS-12s.

For a hybrid restoration subnetwork 40, and in the maxflow phase, the present invention first routes OC-12C failed working capacity over OC-12 spare links. Then, the max flow phase routes OC-3C, failed working capacity, over OC-12 and OC-3 spare links. Next, the present embodiment routes STS-1 failed working links over OC-12, OC-3 and STS-1 spare links. Finally, the max flow phase routes DS3 failed working links over OC-12, OC-3, STS-1, and DS3 spare links. In the connect phase, the restoration subnetwork of the present invention responds to hybrid network in a manner so that tandem nodes get instructions to cross-connect more than one kind of traffic. FIGURE 34 relates to the property of the present invention of assigning different priorities for working paths, and different qualities for spare links. The present embodiment of the invention includes 32 levels of priority for working paths; priority configurations occur at origin node 42, for example. Moreover, the preferred embodiment provides four levels of quality for spare links, such as the following. A SONET 1 for N protected spare link on a span that has no failed links has the highest quality. The next highest quality is a SONET 1 for N protect port on a span that has no failed links. The next highest quality is a SONET 1 for N protected port on the span that has a failed link. The lowest quality is a SONET 1-for-N protect port on a span that has a failed link.

With this configuration, different priorities relate to working paths, and different qualities for spare links. At some stages of employing the present process, the feature of priority working paths and different quality spare links for some uses of the present process, it is possible to simplify the different levels of priority and different levels of quality into simply high and low. For example, high priority working links may be those having priorities 1 through 16, while low priority working links are those having priorities 17 through 32. High quality spares may be, for example, quality 1 spares, low quality spares may be those having qualities 2 through 4. With the varying priority and quality assignments, the present invention may provide a method for restoring traffic through the restoration subnetwork. For example, the present invention may first try to restore high priority failed working links on high-quality spare links, and do this as fast as possible. Next, restoring high-quality failed working links on low-quality spares may occur. Restoring low-priority failed working paths on low-quality spare links occurs next. Finally, restoring low priority failed working paths on high quality spare links. To achieve this functionality, the present invention adds an extra iteration at the end of normal iterations. The extra iteration has the same number of steps as the iteration before it. Its function, however, is to address the priorities for working paths and qualities for spare links. Referring to FIGURE 34, during normal iterations, the present invention will restore high priority working paths over high-quality spare links. During the extra iteration, as the invention restores high- priority working paths over low-quality spare links, then low-priority working paths over low-quality spare links, and finally low-priority working paths over high-quality spare links. This involves running the max flow algorithm additional times.

The network restoration process of the present invention, including the explore, return, and connect messaging phases may be repeated more than once in response to a single failure episode with progressively greater hop count limits. The first set of iterations are confined in restoring only high priority traffic. Subsequent or extra iterations may be used seek to restore whatever remains of lesser priority traffic. This approach give high priority traffic a preference in terms of path length.

FIGUREs 35-37 provide illustrations for describing in more detail how the present invention handles fault isolation. Referring to FIGURE 35, between tandem notes 44 and 46 appear spare link 92. Between custodial nodes 62 and 64 are working link 18 having failure 66 and spare link 196. If a spare link, such as spare link 196, is on a span, such as span 38 that has a failed working link, that spare link has a lower quality than does a spare link, such as spare link 92 on a span that has no failed links. In FIGURE 35, spare link 92 between tandem notes 46 and 48 is part of a span that includes no failed link. In this example, therefore, spare link 92 has a higher quality than does spare link 196. Within each node, a particular order is prescribed for sorting lists of spare ports and lists of paths to restore. This accomplishes both consistent mapping and preferential assignment of highest priority to highest quality restoration paths. Specifically, spare ports are sorted first by type (i.e., bandwidth for STS-12, STS-3), then by quality and thirdly by port label numbers. Paths to be restored are sorted primarily by type and secondarily by an assigned priority value. This quality of a given restoration path is limited by the lowest quality link along the path.

In addition to these sorting orders, a process is performed upon these lists in multiple passes to assign traffic to spare ports while making best use of high capacity, high-quality resources. This includes, for example, stuffing high priority STS-l's onto any STS-12's that are left after all other STS-12 and STS-3 traffic has been assigned.

Rules determine the proper way of handling different priorities of working paths and different qualities of spares in performing the restoration process. In our embodiment of the invention, there may be, for example, 32 priority levels. The working traffic priority may depend on business-related issues, such as who is the customer, how much money did the customer pay for communications service, what is the nature of the traffic. Higher priority working channels are more expensive than are lower priority channels. For example, working are assigned priorities according to these types of considerations. Pre-determined configuration information of this type may be stored in the origin node of the restoration subnetwork. Thus, for every path in the origin node priority information is stored. Although functionally there is no difference between a high priority working path and lower priority working path, though higher priority working paths will have their traffic restored first and lower priority working paths will be restored later.

The present embodiment includes four qualities of spare links. Spare link quality has to do with two factors. A link may either be protected or nonprotected by other protection schemes. In light of the priorities of failed working paths and the quality of spare links, the present invention uses certain rules. The first rule is to attempt to restore the higher priority failed working paths on the highest quality spare links. The next rule is to restore high quality failed working paths on both high quality and low quality spares. The third rule is to restore low priority failed working paths on low quality spares. The last thing to do is to restore low priority working paths over high and low quality spares.

The present invention also it possible for a node to know when it is a custodial node. Because there are no keep-alive messages on working links, however, the custodial node does not know on what span the failed link resides. Thus, referring to FIGURE 36, custodial node 64 knows that custodial node 62 is on the other end of spare link 196. The difficulty arises, however, in the ability for custodial nodes 62 and 64 to know that working link 18 having failure 66 and spare link 196 are on the same span, because neither custodial node 62 nor custodial node 64 knows on what span is working link 18.

FIGURE 37 illustrates how the present embodiment overcomes this limitation. Custodial node 64, for example, sends a "I am custodial node", flag in the keep alive messages that it sends on spare links, such as to non-custodial tandem node 46. Also, custodial node 64 and custodial node 62 both send "I am custodial node" flags on spare 196, to each other. In the event that the receiving non- custodial node, such as tandem node 46, is not itself a custodial node, then it may ignore the "I am custodial node", flag. Otherwise, the receiving node determines that the failure is on the link between itself and the custodial node from which the receiving custodial node receives the "I am custodial node" flag.

There may be some limitations associated with this procedure, such as it may be fooled by "glass throughs" or spans that have no spares. However, the worst thing that could happen is that alternate path traffic may be placed on a span that has a failed link, i.e., a lower quality spare.

The present embodiment provides this functionality by the use ofan "I am custodial node" flag that "piggybacks" the keep alive message. Recalling that a custodial node is a node on either side of a failed link, when the custodial node is identified, the "I am custodial node" flag is set. If the flag appears on a spare link, that means that the neighboring link is the custodial node. This means that the node is adjacent to a failure. If the node receiving the flag is also a custodial node, then the spare is on the span that contains the failed link. So, the custodial node that is sending the flag to the noncustodial node, but not getting it back from a non-custodial node a flag, this means that the spare link is not in a failed span.

FIGUREs 38-42 illustrate the restricted re-use feature of the present invention. The present invention also includes a restricted re-use function. A recovered link relates to the feature of restricted re-use. Given a path with a failure in it, a recovered link may exist between two nodes. The recovered link is a good link but is on a path that has failed. FIGURE 38 shows restoration subnetwork 40 that includes origin node 42 on link 18 and through custodial nodes 62 and 64 connects to destination node 48. Failure 66 exists between custodial nodes 62 and 64. The restricted re-use feature of the present invention involves what occurs with recovered links, such as recovered link 224.

With the present invention, there are at least three possible modes of re-use. One mode of reuse is simply no re-use. This prevents the use of recovered links to carry alternate path traffic. Another possible re-use mode is unrestricted re-use, which permits recovery links to carry alternate path traffic in any possible way. Still another re-use mode, and one that the present embodiment provides, is restricted re-use. Restricted re-use permits use of recovered links to carry alternate path traffic, but only the traffic they carry before the failure.

FIGURE 39 illustrates the restricted re-use concept that the present invention employs. Link 18 enters origin node 42 and continues through tandem node 226 on link 228 and 230 through custodial node 64 through recovered link 48.

Restricted re-use includes modifications to the explore and return phases of the present invention wherein the process determines where recovered links are in the network. The process finds the recovered links and sends this information to the origin node. The origin node collects information about where the recovered links are in the network to develop a map of the recovered links in the restoration subnetwork. The tandem nodes send information directly to the origin node via the wide are network about where the re-use links are.

FIGURE 40 through 42 illustrate how the present embodiment achieves restricted re-use. Referring to restoration subnetwork portion 40 in FIGURE 40, origin node 42 connects through tandem node 44 via link 78, to tandem node 46 via link 82, to tandem node 186 via link 84, and to destination node 48 via link 190. Note that between tandem node 46 and tandem node 186 appears failures 66.

To implement restricted re-use in the present embodiment, during the explore and return phases the origin node 42 will acquire a map of recovered links. Thus, as FIGURE 40 shows within origin node 42, recovered links 232, 234, and 236 are stored in origin node 42. This map is created by sending in-band messages, re-use messages, during the explore phase, along recovered links from the custodial nodes to the origin and destination nodes, such as origin node 42 and destination node 48. Thus, as FIGURE 41 illustrates, in the explore phase, reuse messages emanate from tandem node 46 to tandem node 44 and from there to origin node 42. From tandem node 186, the re-use message goes to destination node 48.

In the return phase, such as FIGURE 42 depicts, the destination node sends the information that it has acquired through re-use messages to the origin node by piggybacking it on return messages. Thus, as shown in FIGURE 42, designation node 48 sends on link 192 a return plus re-use message to tandem node 46. In response, tandem node 46 sends a return plus re-use message on link 76 to origin node 42.

With the restricted re-use feature and in the max flow phase, origin node 42 knows about recovered links and "pure" spare links. When the origin node runs the max flow algorithm, the recovered links are thrown in with the pure spare links. When the breadth-first-search is performed, the present invention does not mix recovered links from different failed working paths on the same alternate path.

Another feature of the present invention relates to spare links connected into paths. In the event of spare links being connected into paths, often these paths may have idle signals on them or a test signal. If a spare link has a test signal on it, it is not possible to distinguish it from a working path. In this instance, the present invention avoids using spare links with "working" signals on them In the max flow phase, the origin has discovered what may be thought of as pure spare link. The origin node also receives information about recovered links, which the present invention limits to restricted re-use. In running the max flow algorithm during the max flow phase of the present process, the pure spare and recovered links and used to generate a restoration map of the restoration subnetwork, first irrespective of whether the links are pure, spare or recovered.

Another aspect of the present invention is the path inhibit function. FIGUREs 43 and 44 illustrate the path inhibit features of the present invention. For a variety of reasons, it may be desirable to temporarily disable network restoration protection for a single port on a given node. It may be desirable, later, to turn restoration protection back on again without turning off the entire node. All that is desired, is to turn off one port and then be able to turn it back on again. This may be desirable when maintenance to a particular port is desired. When such maintenance occurs, it is desirable not to have the restoration process of the present invention automatically initiate. The present invention provides a way to turn off subnetwork restoration on a particular port. Thus, as FIGURE 43 shows, origin node 42 includes path 2 to tandem node 44. Note that no link appears between node 42 and 44. This signifies that the restoration process of the present invention is inhibited along path 240 along origin node 42 and tandem node 44. Working path 242, on the other hand, exist between origin node 42 and tandem node 46. Link 76 indicates that the restoration process of the present invention is noninhibited along this path if it is subsequently restored. During the path inhibit function, the process of the present invention inhibits restoration on a path by blocking the restoration process at the beginning of the explore phase. The origin node either does not send out an explore message at all or sends out an explore message that does not request capacity to restore the inhibited path. This is an instruction that goes to the origin node. Thus, during path inhibit, the process of the present invention is to inform origin node 42, for example, to inhibit restoration on a path by sending it a message via the associated wide area network.

Referring to FIGURE 44, therefore, tandem node 46 sends a path inhibit message to origin node 42. Tandem node 46 receives, for example, a TL1 command telling it to temporarily inhibit the restoration process on a port. It sends a message to origin node 42 for that path via wide area network as arrow 246 depicts.

Tandem node 46 sends inhibit path message 246 with knowledge of the Internet protocol address of its source node because it is part of the path verification message. There may be some protocol involved in performing this function. This purpose would be to cover the situation wherein one node fails while the path is inhibited. Another feature of the present invention is that it permits the inhibiting of a node. With the node inhibit function, it is possible to temporarily inhibit the restoration process of the present invention on a given node. This may be done, for example, by a TL1 command. A node continues to send its step-complete messages in this condition. Moreover, the exerciser function operates with the node in this condition. To support the traditional field engineering use of node port test access and path loopback capabilities, the restoration process must be locally disabled so that any test signals and alarm conditions may be asserted without triggering restoration processing. According to this technique as applied to a given path, a port that is commanded into a test access, loopback, or DRA-disabled mode shall notify the origin node of the path to suppress DRA protection along the path. Additional provisions include automatic timeout of the disabled mode and automatic loopback detection/restoration algorithm suppression when a port receives an in-band signal bearing its own local node ID.

Direct node-node communications are accomplished through a dedicated Wide Area Network. This approach bypasses the use of existing in-band and out-of-band call processing signaling and network control links for a significant advantage in speed and simplicity. In addition, the WAN approach offers robustness by diversity.

A triggering mechanism for distributed restoration process applies a validation timer to each of a collection of alarm inputs, keeps a count of the number of validated alarms at any point in time, and generates a trigger output whenever the count exceeds a preset threshold value. This approach reduces false or premature DRA triggering and gives automatic protect switching a chance to restore individual link failures. It also allows for localizing tuning of trigger sensitivity based on quantity and coincidence of multiple alarms.

The preferred embodiment provides a step Completion Timer in Synchronous DRA. For each DRA process initiated within a network node, logic is provided for automatically terminating the local DRA process whenever step completion messages are not received within a certain period of time as monitored by a failsafe timer. Other causes for ending the process are loss of keep alive signals through an Inter-node WAN link, normal completion of final DRA iteration, consumption of all available spare ports, or an operation support system override command.

Another aspect of the present invention is a method for Handling Staggered Failure Events in DRA. In a protected subnetwork, an initial link failure, or a set of nearly simultaneous failures, trigger a sequence of DRA processing phases involving message flow through the network. Other cuts that occur during messaging may similarly start restoration processing and create confusion and unmanageable contentions for spare resources. The present technique offers an improvement over known methods. In particular, during explore and return messaging phases, any subsequent cuts that occur are "queued" until the next Explore phase. Furthermore, in a multiple iteration approach,

Explore messaging for new cuts is withheld while a final Explore/Return/Connect iteration occurs in response to a previous cut. These late-breaking held over cuts effectively result in a new, separate invocation of the DPA process.

The present invention includes failure notification messages that include information about the software revision and hop count table contents that are presumed to be equivalent among all nodes. Any nodes that receive such messages and find that the local software revision or hop count table contents disagree with those of the incoming failure notification message shall render themselves ineligible to perform further DRA processing. However, a node that notices a mismatch and disable DPA locally will still continue to propagate subsequent failure notification messages. The present invention provides a way to Audit restoration process data within nodes that include asserting and verifying the contents of data tables within all of the nodes in a restoration- protected network. In particular, such data may contain provisioned values such as node id, WAN addresses, hop count sequence table, and defect threshold. The method includes having the operations support system disable the restoration process nodes, write and verify provisionable data contents at each node, then re-enabling the restoration process when all nodes have correct data tables.

In a data transport network that uses a distributed restoration approach, a failure simulation can be executed within the network without disrupting normal traffic. This process includes an initial broadcast of a description of the failure scenario, modified DRA messages that indicate they are "exercise only" messages, and logic within the nodes that allows the exercise to be aborted if a real failure event occurs during the simulation. Another aspect of the present invention is the ability to coordinate with other restoration processes such as, for example, the RTR restoration system. With the present invention, this becomes a challenge because the port that is protected by the restoration process of the present invention is often also protected by other network restoration algorithms. Another aspect of the present invention is the exerciser function. The exerciser function for the restoration process of the present invention has two purposes, one is a sanity check to make sure that the restoration process is operating properly. The other is an exercise for capacity planning to determine what the restoration process would do in the event of a link failure. With the present invention, the exerciser function operates the same software as does the restoration process during subnetwork restoration, but with one exception. During the exerciser function, connections are not made. Thus, when it comes time to make a connection, the connection is just not made.

With the exerciser function, essentially the same reports occur as would occur in the event of a link failure. Unfortunately, because of restrictions to inband signaling, there are some messages that may not be exchanged during exercise that would be exchanged during a real event. For that reason, during the exercise function it is necessary to provide the information that is in these untransmittable messages. However, this permits the desired exerciser function.

Another aspect of the present invention is a dropdead timer and emergency shut down. The drop-dead timer and emergency shut down protect against bugs or defects in the software. If the restoration process of the present invention malfunctions due to a software problem, and the instructions become bound and aloof, it is necessary to free the restoration subnetwork. The dropdead timer and emergency shut down provide these features. The drop-dead timer is actuated in the event that a certain maximum allowed amount of time in the restoration process occurs. By establishing a maximum operational time the restoration network can operate for 30 seconds, for example, but no more. If the 30 second point occurs, the restoration process turns off. An emergency shut down is similar to a drop-dead timer, but is manually initiated. For example, with the present invention, it is possible to enter a TL1 command to shut down the restoration process. The emergency shut down feature, therefore, provides another degree of protection to compliment the drop dead timer.

Out-of-band signaling permits messages to be delivered over any communication channel that is available. For this purpose, the present invention uses a restoration process wide area network. For purposes of the present invention, several messages get sent out of band. These include the explore message, the return message, the connect message, the step complete message, as well as a message known as the exercise message which has to do with an exerciser feature of the present invention. The wide area network of the present invention operates under the TCP/TP protocol, but other protocols and other wide area networks may be employed. In order to use the wide area network in practicing the present invention, there is the need for us to obtain access to the network. For the present invention, access to the wide area network is through two local area network Ethernet ports. The two Ethernet ports permit communication with the wide area network. In the present embodiment of the invention, the Ethernet is half duplex, in the sense that the restoration subnetwork sends data in one direction on one Ethernet while information flows to the restoration subnetwork in the other direction on the other Ethernet port. The wide area network of the present invention includes a backbone which provides the high bandwidth portion of the wide area network. The backbone includes the same network that the restoration subnetwork protects. Thus, the failure in the restoration subnetwork could potentially cut the wide area network. This may make it more fragile. Accordingly, there may be more attractive wide area networks to use with the present invention. For example, it may be possible to use spare capacity as the wide area network. In other words, there may be spare capacity in the network which could be used to build the wide area network itself. This may provide the necessary signal flows to the above-mentioned types of messages. With the present invention, making connections through the wide area network is done automatically. For the cross-connects of the present invention, there is a control system that includes a number of computers within the cross-connect switch. The crossconnect may include possibly hundreds of computers. These computers connect in the hierarchy in three levels in the present embodiment. The computers that perform processor-intensive operations appear at the bottom layer or layer 3. Another layer of computers may control, for example, a shelf of cards. These computers occupy layer 2. The layer 1 computers control the layer 2 computers. The computers at layer 1 perform the instructions of the restoration process of the present invention. This computer maybe centralized in the specific shelf where all layer 1 computers are in one place together with the computer executing the restoration process instructions. Because the computer performing the restoration process of the present invention is a layer 1 computer, it is not possible for the computer itself to send in-band messages. If there is the desire to send an in-band message, that message is sent via a layer 3 computer. This is because the layer 3 computer controls the local card that includes the cable to which it connects. Accordingly, in-band messages are generally sent and received by layer 2 and/or layer 3 computers, and are not sent by layer 1 computers, such as the one operating the restoration instructions for the process of the present invention. Fault isolation also occurs at layer 2 and layer 3 computers within the cross-connects. This is because fault isolation involves changing the signals in the optical fibers. This must be done by machines at lower layers. Moreover, a port, which could be a DS3 port or a SONET port, has a state in the lower layer processors keep track of the port state. In essence, therefore, there is a division of labor between layer 2 and 3 computers and the layer 1 computer performing the instructions for the restoration process of the present invention. With reference to FIGURE 45, an end to end path in a telecommunications network provisioned with a distributed restoration algorithm is shown to include an origin node and a destination node, represented by O and D, respectively. Interconnecting the origin node and destination node are a number of nodes, for example the intermediate nodes N1-N5. Each of these nodes in fact is a digital cross-connect switch such as the 1633-SX switch made by the Alcatel

Company. Each of these switches has a number of ports to which are connected a number of links for interconnecting each switch to other switches of the network. For ease of explanation and illustration, as shown in FIGURE 45, each adjacent pair of nodes is connected by a span or link such as 302, 304, 306, 308, 310 and 312. As is well known, each span can have a number of links and each adjacent pair of nodes may in fact have a number of interconnected spans. Also for ease of discussion, no other nodes of the telecommunications network are shown to be connected to the path of FIGURE 45. As illustrated in FIGURE 45, a fault has occurred in span 306. Such fault may be for example a cut in which one or more links of the span have been cut, or in the worst case scenario, the whole span has been cut. As is well known, when a fault occurs, the nodes bracketing or sandwiching the fault 314 are the first nodes to receive an alarm signal, which then is propagated by those nodes to nodes downstream thereof. Thus, as soon as fault 314 occurs at span 306, custodial nodes N2 and N3 each receive an alarm. Nodes N2 and N3 would in turn propagate the received alarm to nodes downstream thereof such as for example nodes Nl and the origin node for custodial node N2, and nodes N4, N5 and the destination node for custodial node N3. With fault 314 at span 306, the communicative path of FIGURE 45 becomes non-functioning.

This is despite the fact that there is only one fault, namely fault 314 for the whole path. Putting it differently, there is only one span, namely span 306, that has malfunctioned. Yet data from the origin node can no longer be routed to the destination node in the path shown in FIGURE 1. This is so in spite the fact that span or links 302, 304, 308, 310 and 312 each remain operational. One objective of the present invention, as noted above, is to be able to utilize the functioning spans or links of a failed path so that there may be a better utilization of the available resources of the telecommunications network.

To achieve this end, the network of the instant invention is provisioned with the ability, in its distributed restoration algorithm, for the custodial nodes of a fault to send out a message that informs nodes downstream thereof of any portions of the failed path that remain intact and functional. The first step of the inventive scheme is illustrated in FIGURE 46. As shown, a "reuse" message is sent from each of nodes N2 and N3 to their respective adjacent nodes Nl and N4. The reuse message propagated from N2 is designated 316 while the reuse message propagated fromN3 is designated 318. In particular, as shown in FIGURE 49, the reuse message is shown to include an identifier field 320 to which an identifier, represented by R, has been added to designate the message as being a reuse message. The message of FIGURE 49 further includes a variable length route information field 322 to which the identification ID for each node can be added. Other fields of the FIGURE 49 message not germane to the discussion of the instant invention are left blank and are not shown in the messages shown in FIGUREs 46-48.

Returning to FIGURE 46, it can be seen that reuse message 316 has in its route information field the node ID of node N2. On the other hand, reuse message 318 has in its route information field the node ID of node N3.

In receipt of reuse message 316, node Nl would append to the route information field its own node ID, before propagating the reuse message onward to the origin node, by way of link 302. Similarly, upon receipt of the reuse message 318, node N4 would append its own node ID to the route information field of reuse message 318, before propagating the reuse message 318 to node N5 by way of link 310. See FIGURE 47. Thus, as the reuse message, be it 316 or 318, gets propagated from one node to other nodes downstream thereof, additional node IDs are appended to the message, until the message gets to the end node of the path, for example the origin node shown in FIGURE 47. At that point, the origin node reads from the route information field of reuse message 316 to find out what intact portions there are of the failed path. As shown in FIGURE 47, origin node can readily ascertain from reuse message 316 that nodes N2 and Nl are the nodes that have forwarded the reuse message. Therefore, the span or links interconnecting those nodes, as well as the span or link that interconnects node Nl to itself are operating properly. Thus, as far as origin node 0 is concerned, the fault that causes the path to fail occurs somewhere beyond node N2, and therefore the elements before node N2 remain usable and can be restricted for use by an alternate route for rerouting the traffic that had been disrupted by fault 314.

Also shown in FIGURE 47 is the reuse message 318, as sent by custodial node N3. As shown, reuse message 318 has been propagated by node N4 to node N5. As seen by node N5, reuse message 318 has in its route information field node N3 and node N4. Therefore, node N5 knows that the path connecting it to node N3 remains good. Node N5 then appends its own node ID to the route information field of reuse message 318 before propagating it, via link 312, to the destination node.

As shown in FIGURE 48, the destination node is now in receipt of reuse message 318. From the route information field of reuse message 318, the destination node can ascertain that fault 314 occurs beyond node N3 and that links 308, 310 and 312 interconnecting nodes N3, N4, N5 remain usable. For the purpose of the instant invention, the conveying of information between the intermediate nodes, such as between nodes N3, N4, N5 and the destination node is done by means of in-band massaging between those nodes. Similarly, the propagation of the reuse message between nodes N2, Nl and the origin nodes are done by in-band massaging.

Once in receipt of reuse message 318, the destination node repackages that reuse message into another "reuse" message 320 and transmits that message to the origin node by means of a wide area network (WAN) massaging connection 322. In receipt of the reuse message 320, the origin node becomes aware of all intact portions of the failed path shown in FIGURE 48. It can then formulate an alternate path by using, for example, the spare link 324 that interconnects node Nl to N5. Thus, for the exemplar embodiment shown in FIGURE 48, the alternate restoration path is able to use links 302 and 312 of the failed path for rerouting the traffic from the origin node to the destination node. Of course, other alternate route(s) interconnecting the origin node to the destination node could also be used.

In order to enhance the restoration of the failed path once fault 314 is fixed, the useful links of the failed path of FIGURE 48 are, for the most part, restricted for routing information from the origin node to the destination node. In other words, in addition to links 302 and 312, intact links 304, 308 and 310 are reserved for the use of the origin node and the destination node for routing data therebetween.

Even though FIGURE 48 illustrates that a different reuse message 320 is sent by the destination node to the origin node, it should be appreciated that a reuse message can also be sent from the origin node to the destination node 322 to inform the destination node of links or spans of the failed path that remain functional. In fact, both end nodes of the failed path can inform each other of intact portions of the failed path, if needed.

The exemplar telecommunications network of the instant invention, as shown in FIGURE 50, comprises a number of nodes 6302-324 each connected to adjacent nodes by at least one working link and one spare link. For example, node 6302 is connected to node 6304 by means of a working link 2- 4W and a spare link 2-4S. Similarly, node 6304 is connect to node 6306 by a working link 4-6W and a spare link 4-6S. For the sake of simplicity, only the specific links connecting nodes 6302-6304, 6304-6306 and 6302-6310 are appropriately numbered in FIGURE 50. But it should be noted that the working and spare links connecting adjacent nodes can be similarly designated.

For the telecommunications network of FIGURE 50, it is assumed that all of the nodes of the network are provisioned with a distributed restoration algorithm (DRA), even though in practice oftentimes only one or more portions of the telecommunications network are provisioned for distributed restoration. In those instances, those portions of the network are referenced as dynamic transmission network restoration (DTNR) domains.

Also shown in FIGURE 50 is an operation support system (OSS) 6326. OSS 6326 is where the network management monitors the overall operation of the network. In other words, it is at OSS 6326 that an overall view, or map, of the layout of each node within the network is provided. OSS 6326 has a central processor 6328 and a memory 6330 into which data retrieved from the various nodes are stored. Memory 6330 may include both a working memory and a database store. An interface unit, not shown, is also provided in OSS 6326 for interfacing with the various nodes. As shown in FIGURE 50, for the sake of simplicity, only nodes 6302, 6304, 6306, and 6308 are shown to be connected to OSS 6326. Given the interconnections between OSS 6326 and the nodes of the network, the goings on within each of the nodes of the network is monitored by OSS 6326.

Each of nodes 6302-6324 of the network comprises a digital cross-connect switch such as the 1633-SX broadband cross-connect switch made by the Alcatel Network System company. Two of such adjacently connected switches are shown in FIGURE 51. The FIGURE 51 switches may represent any two adjacent switches shown in the FIGURE 50 network such as for example nodes 6304 and 6306 thereof. As shown, each of the switches has a number of access/egress ports 6332, 6334 that are shown to be multiplexed to a line terminating equipment (LTE) 6336, 6338. LTEs 6336 and 6338 are SONET equipment having a detector residing therein for detecting any failure of the links between the various digital cross-connect switches. Again, for the sake of simplicity, such LTE is not shown to be sandwiched between nodes 6334 and 6336, as detection circuits for interpreting whether a communication failure has occurred may also be incorporated within the respective working cards 6340a, 6340b of node 6304 and 6342a and 6342b of node 6306.

As shown in FIGURE 51, each of the digital cross-connect switches has two working links 6344a and 6344b communicatively connecting node 6304 and node 6306, by means of the respective working interface cards 6340a, 6340b and 6342a, 6342b. Also shown connecting node 6304 and node 6306 are a pair of spare links 6346a and 6346b, which are connected to the spare link interface cards 6348a, 6348b and 6350a, 6350b of node 6304 and node 6306, respectively. For the FIGURE 51 embodiment, assume that each of working links 6344a, 6344b and spare links 6346a, 6346b is apart of a logical span 6352. Further note that even though only four links are shown to connect node 6304 to node 6306, in actuality, adjacent nodes may be connected by more or less links. Likewise, even though only four links are shown to be a part of span 6352, in actuality, a span that connects two adjacent nodes may in fact have a greater number of links. For the instant discussion, assume that working links 6344a and 6344b correspond to the working link 4-6W of FIGURE 50 while the spare links 6346a and 6346b of FIGURE 51 correspond to the spare link 4-6S of FIGURE 50. For the purpose of this aspect of the instant invention, each of the links shown in FIGURE 51 is presumed to be a conventional optical carrier OC-12 fiber or is a link embedded within a higher order (i.e., OC-48 or OC-192) fiber.

Focusing onto node 6304 for the time being, note that each of the interfacing card, or boards, of that digital cross-connect switch such as 6340a, 6340b, 6348a and 6348b are connected to a number of STS-1 ports 6352 for transmission to SONET LTE 6336. Although not shown, an intelligence such as a processor residing in each of the digital cross-connect switches controls the routing and operation of the various interfacing boards and ports. Also not shown but present in each of the digital cross-connect switches is a database storage for storing a map which identifies the various sender nodes, chooser nodes and addresses, which will be discussed later. The working boards 6342a, 6342b and the spare boards 6350a, 6350b are likewise connected to the access/egress ports 6354 in node 6306. Further shown in FIGURE 51 are non-DRA between adjacent nodes 6304 and 6306.

For the instant invention, the access/egress ports such as 6332 and 6334 send their respective port numbers through the matrix in each of the digital cross-connects to its adjacent nodes. Thus, for the exemplar interconnected adjacent nodes 6304 and 6306, ports 6352a and 6352b of node 6304 are connected to ports 6354a and 6354c of node 6 by means of working link 6344a. Similarly, ports 6352e and 6352f are interconnected to ports 6354e and 6354f of node 6306 by way of spare links 6346a and 6346b, respectively. Thus, if node 6304 were to transmit a signal using spare link 6346a to node 6306, it will be transmitting such a message from its port 6352e to spare card 6348a, and then onto spare link 6346a, so that the message is received at spare card 6350a of a conventional optical carrier OC-12 fiber or is a link embedded within a higher order (i.e., OC-48 or OC-192) fiber.

For the instant invention, the access/egress ports such as 6332 and 6334 send their respective port numbers through the matrix in each of the digital cross-connects to its adjacent nodes. Thus, for the exemplar interconnected adjacent nodes 6304 and 6306, ports 6352a and 6352b of node 6304 are connected to ports 6354a and 6354c of node 6306 by means of working link 6344a. Similarly, ports 6352e and 6352f are interconnected to ports 6354e and 6354f of node 6306 by way of spare links 6346a and 6346b, respectively. Thus, if node 6304 were to transmit a signal using spare link 6346a to node 6306, it will be transmitting such a message from its port 6352e to spare card 6348a, and then onto spare link 6346a, so that the message is received at spare card 6350a of node 6306 and then routed to the receiving port 6354e of node 6306. Thus, as long as each of the working links and spare links interconnecting a pair of adjacent nodes, such as for example nodes 6304 and 6306 are operational, when a message is sent between those nodes, the information relating to the respective transmit and receiving ports can be collected by the OSS 6326 (FIGURE 50) so that a record can be collected of the various ports that interconnect any two adjacent nodes. For the instant invention, the inventors have seized upon the idea that a topology, or map, of the available spare capacity of the network, in the form of the available spare links that interconnect the nodes, can be generated from stored data that is representative of the different port numbers of the various nodes to which spare links are connected. In other words, if a message transmitted by one node to its adjacent node is able to provide OSS 6326 a number of parameters which include for example the ID of the transmit node, the respective IP (internal protocol) addresses of the transmit and receiving ports of the node and the port number from which the message is transmitted from the node, the OSS can ascertain, from similar messages that are being exchanged between adjacent nodes on spare links connecting those adjacent nodes, an overall picture of the spare capacity of the network.

Simply put, if each of the digital cross-connect switches in the DRA provisioned network knows what port number and the node that it is connected to by its spare link, then that node knows how to reroute traffic if it detects a failure in one of its working links. And by collecting the information relating to each of the nodes of the network, the OSS 6326 is able to obtain an overall view of all of the available spare links that interconnect the various nodes. As a consequence, when a failure occurs at a given working link, OSS 6326 can send to the custodial nodes of the failed link a map of the spare capacity of the network, so that whichever custodial node designated as the sender or origin node can then use that map of the spare capacity of the network to begin the restoration process by finding an alternate route for rerouting the disrupted traffic.

The structure of the special message to be used for continuously monitoring the available spare capacity of the network is shown in FIGURE 52. For the instant invention, this message is referred to as a keep alive message. As shown, this keep alive message has a number of fields. Field 6356 has an 8 bit message field. For the FIGURE 52 message, the 8 bits of data can be configured to represent the keep alive message so that each node in receipt of the message will recognize that it is a keep alive message for updating the availability status of the spare link from which the message is received. OSS 6326, on the other hand, upon receipt of a keep alive message, would group it with all the other keep alive messages received from the different nodes for mapping the spare capacity of the network.

The next field of the message of FIGURE 52 is field 6358, which is an 8 bit field that contains the software revision number of the DRA being used in the network. The next field is 6360, which is an 8 bit field containing the node identifier of the transmitting node. Field 6362 is a 16 bit field that contains the port number of the transmitting node from which the keep alive message is sent.

The next field of the message is field 6364. This is a 632 bit field that contains the IP address of the DS3 port on the node that is used for half duplex incoming messages. The IP address of the DS3 port of the node that is used for half-duplex outgoing messages is contained in the 632 bit field 66. Field 6368 is a 1 bit field that, when set, indicates to the receiving node that the message is sent from a custodial node for a failure. In other words, when there is a failure, the custodial node of the failed link will send out a keep alive message that informs nodes downstream thereof that the keep alive message is being sent from a custodial node since a failure has occurred, and a restoration process will proceed.

The last field of the keep alive message is field 6370. It has 7 bits and is reserved for future usage.

In operation, before any failure is detected, keep alive messages such as that shown in FIGURE 52 are exchanged on the spare links between adjacent nodes continuously. By the exchange of these keep alive messages, the network is able to keep a tab of the various available and functional spare links and also identify the port number of each node from where each spare link outputs a keep alive message, as well as the port number of the adjacent node to which the spare link is connected and to which the keep alive message is received. By collecting the data that is contained in each of the keep alive messages, a record is kept of the various nodes, the port numbers, the incoming and outgoing IP addresses of the various spare links that are available in the network. And from these collected data, a topology of the available spare capacity of the network can be generated, by either the OSS 6326, or by each of the nodes, which can have the collected information downloaded thereto for storage. In any event, a map of the available spare links of the network is available, so that when a failure does occur, the custodial nodes of the failure could retrieve the up-to-date map of the spare capacity of the network, and based on that, be able to find the most efficient alternate route for rerouting the disrupted traffic.

Given that the instant invention relates to a distributed restoration process, it should be noted that an OSS is not necessary for storing the topology of the spare capacity of the network, as each of the digital cross-connect switches of the network knows what port number and the nodes that it is connected to by its spare links. Thus, when a failure occurs, each of the nodes will continue to send the keep alive message, as the origin node that is responsible for restoration can build the entire topology of the available spare links by retrieving the different keep alive messages from the various nodes. Putting it differently, an origin node, in attempting to determine the available spare links, only needs to take the sum of all of the keep alive messages since each node that has at least one spare link will send a keep alive message to the origin node. And, by retrieving the ID of the node and the port numbers of the node to which spare links are connected, the spare capacity of the network can be ascertained. As a consequence, the map of the spare link topology becomes available in a distributed matter to the origin node in the instant invention DRA provisioned network.

As previously stated, in the DRA network the C-bit is used to exchange keep alive (KA) messages on spare links, although it may be employed on links carrying a payload. A link has two ports, one on each end of the link. KA messages gives each port information about the other port. A node has one or more ports to which the links between nodes are connected. In the conventional configuration, the node does not know nor needs to know what other node the links that are connected to it are. For example, as shown in FIGURE 61 (See also FIGURE 51), node 100 has ports 10, 15 and 20 and node 200 has ports 30, 35 and 40. Node 100 through port 20 is connected to port 30 of node 200 via a link. A link of this nature may be represented by the following short hand notation (node 100, port 20) to (node 200, port 30). In a conventional configuration, node 100 and node 200 do not know that they are connected to each other.

In one aspect of the present invention, the KA messages, as previously stated, gives each port and in turn each node information about the other port. The KA messages are exchanged during normal operation. The information contained within the KA messages and specifically carried by or embedded in the C-bit, although other segments within the DS3 signal may carry the information, is utilized during restoration of a failed link or span.

Generally in telecommunications so-called performance monitoring is done, wherein one port or a node looks at its incoming signal, monitors the quality of that signal, then it reports to a higher authority what the quality of that signal is. The corresponding port at the other end of the link does the same type of monitoring. The information in the KA message can be expanded to contain far-end "Quality of Service" (QoS) information.

As stated, the QoS information is a measure of the quality of the signal being received at each port of the link. This information can include, but is not limited to, received errored seconds, received severely errored seconds, and received Loss of Signal (LoS). The QoS information can be reported over any predetermined time period. The present aspect of the invention contemplates the time intervals to be one of the last 15 minutes, last hour, and last day. The time over which KA messages and hence the QoS information is sent could be continuously. The additional information can be used to assign a quality value to the link, usually the spare link. The better the QoS, the higher the quality of the link. Since both ports have the same information, both ports assign the link the same quality value. A quality value is associated with the transmission of data from one port to another in both directions and it may be the case that the quality value is better in one direction than the other. For instance, in FIGURE 61 the quality value may be equal to 3 for the transmission of information from (node 100, port 20) to (node 200, port 30), but exhibit a quality value equal to 5 for the information transmitted from (node 200, port 30) to (node 100, port 20), assuming a scale from 1 to 10 is used with one being the best and 10 the worst. Therefore, the link between these ports could be assigned an averaged quality value of 4. Alternately, the (node 100, port 20) to (node 200, port 30) could be assigned a value of 5 if it is decided that the lowest of the two QoS values should be assigned to the link. With this additional information, the assignment of data to a specific link may be modified based on several criteria. For example, during a restoration event the distributed restoration algorithm could determine that the data that has the highest priority should be placed on the link with the best QoS and the data with the second highest priority should be placed on the link with the second best QoS.

For instance, the banking or stock brokerage information in the payload may be given the highest priority whereas system information, called overhead, may be given the lowest priority. Depending on the type of disruption that has occurred the system information may be most critical and then would receive the highest priority. The combinations are numerous and will not be explored further at this time.

Now turning back to when a fault occurs, with reference to prior art FIGURE 53, a Digital Service 3 (DS3) path that connects node 7301 to node 7306 of a distributed restoration domain of a telecommunications network is shown. For the sake of simplicity, no other nodes of the network, or the domain, are shown. As is well known, in a DRA network, when a fault occurs at a link interconnecting two adjacent nodes, an alarm is generated and sent to each of the adjacent nodes. Such a fault, or a malfunctioned link, is shown to have occurred as a failure between nodes 7303 and 7304 in FIGURE 53. This failure may be due to, for example, a loss of signal (LOS), a loss of frame (LOF), or a loss of pointer (LOP) in the signal traversing between nodes 7303 and 7304. For the discussion of the instant invention, assume such an alarm signal is an alarm indication signal (AIS).

In the prior art, each of the nodes of a distributed restoration network, or domain, is provisioned to follow the standard set forth in the Bellcore document TR-NWT-00170 which mandates that each node downstream of the custodial nodes, such as nodes 7303 and 7304, upon receipt of the AIS signal, in turn should propagate the signal to nodes downstream thereof. Thus, in the illustrated FIGURE 53, upon receipt of the AIS signal, node 7303 propagates the AIS signal to node 7302, which in turn propagates it to node 7301, which in turn propagates it along the DS3 path to nodes downstream thereof. The same flow of the AIS signal received by node 7304 occurs with respect to node 7304, node 7305 and node 7306. For the FIGURE 53 embodiment, assume node 7301 and node 7306 are access/egress ports each communicatively interconnecting the distributed restoration domain to other parts of the telecommunications network, or other networks in the case where the distributed restoration domain is not any part of any network.

The problem with the prior art distributed restoration domain is that since most, if not all, of the nodes of the domain will eventually receive the AIS signal, it is quite difficult, if not impossible, for the nodes to determine which are the true custodial nodes, i.e. the nodes that bracket or sandwich the fault. Thus, even though the management of the network recognizes readily that a fault has occurred at a certain path, it nonetheless could not isolate the precise location where the fault occurred.

An aspect of the present invention is applicable to networks incorporating digital cross- connect systems generally, and particularly to networks incorporating broadband 1633-SX digital cross-connect switches. FIGURE 54 illustrates one embodiment of the present invention in which a fault, or a malfunctioned link, could be readily isolated. For the FIGURE 54 illustration, note that each of the nodes 7301-7306 is connected to an operations support system (OSS) 7310, which monitors the operational status of each of the nodes. Similar to the scenario shown in FIGURE 53, a fault is presupposed to have occurred between node 7303 and node 7304. At the time of the link failure, node 7303 and node 7304, each of which being a digital cross connect switch such as that shown in FIGURE 59, detects either a LOS defect, a LOF defect, or an AIS defect signal, each defined by the American National Standard Institute (ANSI) standard Tl .231. For ease of discussion, assume that an AIS signal is detected. When a switch, such as node 7303 or node 7304, detects an AIS signal, normally it would propagate or pass the AIS signal downstream to the next switch along the path such as node 7302 or node 7305, respectively.

In the embodiment of the present invention shown in FIGURE 54, each of the switches or nodes is an intelligent network element provisioned with the appropriate hardware to convert or modify a received AIS signal into a modified AIS signal, before propagating such non-alarm signal to nodes downstream thereof. For the FIGURE 54 embodiment, such non-alarm signal is a DS3 idle signal. Thus, with respect to node 7303 and node 7304, note that when each of those nodes receives the AIS signal, it converts the received AIS signal into an idle signal and propagates the idle signal to node(s) downstream from its output port. For the present invention embodiment, the DS3 idle signal contains an embedded message on the C-bit maintenance channel that identifies the presence of a fault within the distributed restoration domain. Upon receipt of an idle signal converted from an AIS signal, each of the downstream nodes either transmits or propagates the idle signal to nodes downstream therefrom. Thus, node 7302. passes the idle signal received at its input port, via its output port, to node 7301. Similarly, node 7305, on receiving an idle signal from node 7304 at its input port, would transmit the idle signal from its output port to node 7306. This process is repeated ad infinitum until the idle signal reaches the access/egress nodes that interconnect the distributed restoration domain to the rest of the network.

Thus, for the exemplar embodiment of FIGURE 54, given that only node 7303 and node 7304 are in receipt ofan AIS signal, the management of the distributed restoration domain, per monitoring of the domain by OSS 7310, can readily ascertain that the fault occurred between node 7303 and node 7304, and that traffic should be rerouted away from the malfunctioned link connecting nodes 7303 and 7304. As for the network outside of the distributed restoration domain, since such network is not capable of distributedly restoring disrupted traffic and is also in most instances not controlled by the management of the domain, the idle signal has to be reconverted back into the AIS signal so that, as far as the equipment positioned along the paths of the outside network is concerned, an alarm has occurred somewhere within the distributed restoration domain and that appropriate action needs to be taken. To achieve this end, at each of the access/egress nodes of the domain, there is further provisioned the functionality of reconverting an idle signal received at its input port into an AIS signal to be sent via its output port to the nodes downstream thereof in the network outside of the distributed restoration domain. With the conversion of the idle signal back to the AIS signal at the access/egress nodes, customers or equipment outside of the distributed restoration domain continue to receive standards compliant AIS signals.

The process by which an alarm signal is converted into a non-alarm signal, i.e. the AIS signal into an idle signal, is explained herein with reference to FIGURE 55, which shows a DS3 frame structure, in accordance with the format promulgated under ANSI Standard Tl .107-95 for example. In particular, a DS3 signal is partitioned into M-frames of 4760 bits each. The M-frames each are divided into 7 M-subframes each having 680 bits. Each M-subframe in turn is further divided into 8 blocks of 85 bits with 84 of the 85 bits available for payload and one bit used for frame overhead. Thus, there are 56 frame overhead bits in a M-frame. These are divided into a number of different channels: an M-frame alignment channel (Ml, M2, and M3), a M-subframe alignment channel (FI, F2, F3 and F4), a P-bit channel (PI and P2), an X-bit channel (XI and X2), and a C-bit channel (Cl, C2 and C3).

The M-frame alignment channel signal is used to locate all 7 M-frames. The M-subframe alignment channel signal is used to identify all frame overhead bit positions. The P-bit channel is used for performance monitoring, with bits PI and P2 being set to 11 or 00. The C-bit channel bit (Cl, C2 and C3) positions are reserved for applications specific uses. According to the ANSI T 1.107- 95 Standard, the C-bit channel can be employed to denote the presence or absence of stuffed bits. Thus, if the 3 C-bits (Cl, C2, and C3) in the M-subframes are set to 1, stuffing occurs. If those C-bits are set to 0, there is no stuffing. Also, a majority vote of the three stuffing bits in each of the M- subframes is used for the determination. Additional description of the various bits, the M-subframes and M-frame of a DS3 signal can be gleaned from the aforenoted Tl .107-95 Standard.

To convert an AIS signal into an idle signal with an embedded message, the inventors seize upon the fact that the 3 C-bits in M-subframe 5 are typically used and set to 1 but are allowed by ANSI T1.107 for use as a datalink. Thus, by changing at least one of those C-bits in M-subframe 5, the digital cross-connect switch, i.e. the node, can transmit an embedded message within an otherwise standard idle signal. For example, when AIS signal is detected at the node, due to a fault or malfunction having occurred at a link adjacent to the node, according to the present invention, the node converts the AIS signal to an idle signal by blocking the AIS received so that it does not pass through the node and instead transmitting OS3 idle signal as defined by ANSI Tl.107-95 in its place. At the same time, the node begins transmitting an embedded message by changing the 3 C-bits in M- subframe 5 of the idle signal. Thus, what is output from the node is an idle signal with a changed C- bit.

To the nodes downstream of the custodial nodes, for example nodes 7302 and 7305, the detected incoming idle signal, even though it contains all the conventional attributes of a standard idle AIS signal, nonetheless has an embedded message or change to it due to the change of the state of at least one of the C-bits, so that those nodes are put on notice that that idle signal contains a message not found in a standard signal. And when this idle signal with the changed C-bit is propagated to an access/egress port such as node 7301, in sensing this unconventional idle signal, node 7301 will reconvert the idle signal back into an AIS signal for propagation to the nodes outside the distributed restoration domain. This is in contrast to the access/egress node having received a conventional AIS or idle signal, in which case the same AIS or idle signal is propagated to nodes downstream thereof outside the distributed restoration domain.

Thus, with the present invention, given that any node within the distributed restoration domain, in receipt ofan AIS signal, would convert the same into an idle signal with a changed C- bit(s), there are at most only two nodes within the distributed restoration domain that would detect an AIS signal. These two nodes obviously must be adjacent nodes that are connected by the malfunctioned link from where the alarm signal originated. With that in mind and with the continuous monitoring of all of the nodes of the distributed restoration domain by OSS 7310, the isolation of a fault in the distributed restoration domain is easily accomplished.

If a fault occurs in the network outside of the distributed restoration domain, the AIS signal generated as a result of the fault would enter the distributed restoration domain at any one of its access/egress nodes. These nodes that interconnect the domain to the outside network are provisioned such that any incoming alarm signal is converted to a nonalarm signal to be propagated to the other nodes within the distributed restoration domain. As before, in the case ofan AIS signal, the access/egress node would convert the AIS signal into an idle signal with an embedded message of at least one changed C-bit so that this converted idle signal is routed throughout the distributed restoration domain, until it reaches another access/egress node that interconnects the distributed restoration domain to the outside network at another end of the domain. At which time the second access/egress node, upon sensing the idle plus changed C-bit signal, would reconvert that signal back into an AIS signal and propagate it to nodes downstream thereof outside of the distributed restoration domain. With this conversion and reconversion ofan AIS signal into the distributed restoration domain, the management of the domain becomes aware that a fault has occurred in the telecommunications network, but outside of its domain. FIGURE 56 illustrates another aspect of the present invention in which the nodes of a distributed restoration domain are interconnected by optical fibers. The same discussion with respect to the FIGURE 55 embodiment is equally applicable herein but for the fact that a different type of signal, namely a SONET STS-n type signal, is transmitted among the nodes of the domain and the telecommunications network to which the domain is interconnected. For this type of STS-n signal, in the event of a link failure, the custodial cross-connect switches on either side of the malfunctioned link would detect one of the following conditions: loss of signal (LOS), loss of frame (LOF), an alarm indication signal-line (AIS-L), a path loss of pointer (LOP-P) and a path AIS (AIS-P). All of these various defect signals are referenced in the ANSI T1.231 Standard. As with the asynchronous scenario discussed above with reference to FIGURES 54 and 55, the STS-path AIS signal will propagate throughout the. network when a SONET link fails, and the same process of conversion and reconversion as discussed above takes place in the FIGURE 56 embodiment. But instead of the format shown in FIGURE 55, a STS-n frame format, such as the STS-3 frame shown in FIGURE 57, is used. And instead of the C-bit, for the STS-n type format, the inventors found that the bits that can be manipulated are the Z5 bits in the payload section of the STS- n format, for example the STS-3 format of FIGURE 57. As was done with the DS3 format shown in FIGURE 55, by changing the state of one of the Z5 bits, the custodial nodes can convert the detected AIS signal into an idle signal with a changed Z5 bit. When this changed idle signal reaches the access/egress node of the distributed restoration domain, it in turn is reconverted back into an AIS signal and propagated to nodes outside the domain.

FIGURE 58 is a logical diagram illustrating the various layers of processing that take place in a digital cross-connect switch. For example, an AIS signal is fed to the input port, and specifically a port card 7312a in layer 3, or level 3, that performs the basic lower-level functions such as detecting an incoming signal and whether that signal is an alarm or not. The signal is then provided to a shelf processor 7314a in layer 2 that performs other processing functions. In the event that an idle signal with a C-bit message, i.e. one of the C-bits in one of the subframes having been changed, comes onto port card 12b, the contents of the C-bit message are dropped at the port card level and the message is sent to the shelf processor 7314b at level 2 and then routed to a DRA processor 7316 so that a decision can be made as to which port the message is to be output. The C-bit message is also routed to the administrative processor 7318 which, together with DRA processor 7316, accesses the appropriate database (not shown) to obtain information on the particular port and the card in the cross-connect switch to which the signal is to be ultimately routed, so that the signal is routed to the appropriate input port in the cross-connect switch downstream of the node. The reverse operations occur with an input ofan AIS signal in the FIGURE 58 logical diagram. An exemplar node of the-instant invention is illustrated in FIGURE 59. As shown, a number of transceiver units 7320 are connected to a digital cross-connect switch 7322 by way of demultiplexers 7324 and multiplexers 7326. The transceiver units 7320 each are connected to an alarm processor 7328, a restoration signal sensor 7330 and a restoration signal generator 7332. The alarm processor 7328, restoration signal sensor 7330 and restoration signal generator 7332 are connected to the processor of the node 7334. The operation of the cross-connect switch 7322 is of course controlled by node processor 7334.

Within each transceiver units 7320 there is a signal transceiver detector 7336 which is in communication with a frame receive unit 7338 and a frame transmitter unit 7340. Frame receive unit 7338 in turn is connected to a converter 7342, while frame transmitter unit is connected to a converter 7344. Since each of the transceiver units 7320 contains the same components and operates in the same way, only the operation of one of the transceiver units 7320 is discussed hereinbelow.

In receipt of a signal, signal transceiver detector 7336 determines whether the signal is an alarm signal or other types of signal. If the input signal is indeed an alarm signal, signal transceiver detector 7336 first signals alarm processor 7338 via interface 7336 and then routes the signal to frame receive unit 7338. There the signal is parsed and forwarded to converter 7342 so that the alarm signal is converted into the non-alarm signal, with the state of the appropriate C-bit(s) being changed, if it is a DS3 system. A signal is also forwarded to the restoration signal sensor 7330, to be further transmitted to the node processor 7334.

The non-alarm signal is then forwarded to demultiplexer 7324 and then the digital cross connect switch 7322. With the appropriate determination from node processor 7334, the appropriate port in the node through which the non-alarm signal is to be output to the downstream nodes is selected so that the non-alarm signal is provided to multiplexer 7326 and then to converter 7344, if further conversion is required. The non-alarm signal is then provided to frame transmit unit 7340 and the signal transceiver detector 7336 and provided to the appropriate port for output to downstream nodes. Note that the node shown in FIGURE 59 is also provisioned as an access/egress node so that in the event that a signal input thereto is outside the distributed restoration domain, that signal, if indeed it is an AIS signal, is converted into an idle signal with a changed C-bit message and propagated to the nodes downstream thereof within the distributed restoration domain until it reaches an access/egress node at the other end of the domain. At which time the idle signal with the changed C-bit message is reconverted back into an AIS signal and propagated to the nodes outside the distributed restoration domain. For an in depth discussion of a node and its various units for receiving and transmitting signals therefrom in a SONET environment, the reader is directed to U.S. patent 5,495,471, the disclosure of which being incorporated by reference herein.

FIGURE 60 is a diagram illustrating the respective statuses of the various signals both within the distributed restoration domain of a DS3 system and outside of the domain. Specifically, note that outside of the distributed restoration domain, the types of ports that are being used are both nonDRA and that, insofar as the management of the distributed restoration domain is concerned, no attention needs to be paid to the input and output signals. See the first row of blocks designated 7350. At the DRA access port for an access/egress node, with an input signal being an AIS signal, the signal being output becomes an idle signal with a changed C-bit message. See row 7352a. If the signal to the DRA access input port is an idle signal, however, an idle signal is output from the node. See row 7352b. For the same access node, there should not be any idle plus C-bit message inputting from outside of the distributed restoration domain. Accordingly, there is no output signal from the access node shown in row 7352c of FIGURE 60.

Focus now to the egress side ofan access/egress node of the distributed restoration domain, as shown in rows 7354 of FIGURE 60. There, upon receipt ofan AIS signal from within the domain means, that AIS signal needs to be provided to the nodes outside of the domain, as indicated at row 7354a. This of course means that the node within the distributed restoration domain that is adjacent to the access/egress node likewise would receive an AIS signal, and that the fault should be isolated to the link that connects the access/egress node to its adjacent node within the domain. At row 7354b, note that, similar to row 7354a, if an idle signal is provided to the access/egress from within the domain, an idle signal is output from the access/egress node to the nodes outside the domain. At row 7354c, note that if an idle signal with a changed C-bit message is received at the access/egress node from within the restoration domain, the access/egress node will reconvert this previously converted alarm signal back into an AIS signal, and will propagate the AIS signal to nodes downstream thereof outside of the distributed restoration domain.

Row 7356 of FIGURE 60 illustrates the input and output signals of a node other than an access/egress node within the distributed restoration domain. As shown in row 7356a, if the DRA provisioned node were to receive an AIS signal, then byway of the processing discussed earlier, this AIS signal is converted into an idle signal with a changed C-bit message. On the other hand, if the node were to receive an idle signal, then nothing is to be done, as the same idle signal is output from the node and propagated to nodes downstream thereof. See row 7356b. Similarly, per row 3s56c, note that if an idle signal with a changed C-bit message is received by the node, the same idle signal with the changed C-bit message is sent to nodes downstream thereof. Thus, the only time a node within the distributed restoration domain performs a conversion process is when it receives an alarm signal, such as the AIS signal. Putting it differently, the only time that it converts an AIS signal into an idle signal is when a link connected thereto becomes defective. Thus, by locating the adjacent pair of nodes within the distributed restoration domain each of which has detected an alarm signal, default within the domain is easily located.

Even though the discussion so far deals with the interconnection of nodes in a distributed restoration domain by means of links, it should be appreciated that the instant invention is equally applicable to a distributed restoration domain whose nodes are communicatively interconnected without links. For example, in the case of a restoration domain that operates by using microwave transmission, no physical links are used. Instead, the various nodes are interconnected by microwave transmission, which also has its own distinct format. The packet message for a microwave transmission also includes particular bits that could be altered using the same principle of the instant invention so that, if a given domain of a wireless network is provisioned for distributed restoration, to quickly locate or isolate the fault in the event that a malfunction has occurred, the same altering of some unused bits in the microwave message can change the status of the signal that is being transmitted without affecting the operation of the network.

In the case that in place of a link, it is a node that malfunctions, the present invention is equally applicable insofar as that malfunctioning node also generates an alarm signal that is to be received by nodes adjacent thereto so that those adjacent nodes would convert the alarm signal into a non-alarm signal with an embedded message which is then propagated to nodes further downstream thereof in the distributed restoration domain. Thus, the site at which the fault occurred, be it a link or a node, could be isolated nonetheless. While preferred embodiments of the present invention have been disclosed for purposes of explanation, numerous changes, modifications, variations, substitutions, and equivalents, in whole or in part, should now be apparent to those skilled in the art to which the invention pertains. Accordingly, it is intended that the invention be limited only by the spirit and scope of the hereto appended claims. Inasmuch as the present invention is subject to many variations, modifications and changes in detail, it is intended that all matter described throughout this specification and shown in the accompanying drawings be inteφreted as illustrative only and not in a limiting sense. Accordingly, it is intended that the invention be limited only by the spirit and scope of the appended claims.

Claims

CLAIMS 1. A method for identifying a pair of neighboring nodes in a telecommunications network having at least one distributed restoration sub-network, comprising: constructing a first C-bit keep alive message for a first node in a neighboring node pair connected by a link; embedding the first C-bit keep alive message within the C-bit of a first DS3 signal; determining a quality of service information for the link when looking from the first node to the neighboring node; embedding the quality of service information within the C-bit keep alive message; and transmitting the first DS3 signal from the first node to a second node in the neighboring node pair over the link, wherein the first C-bit keep alive message identifies the first node to the second node and the quality of service information for the link when looking from the first node to the second node.

2. The method according to Claim 1, further comprising: receiving the first DS3 signal at the second node; and processing the first DS3 signal to identify the first node from the first C-bit keep alive message to the second node and identifying the quality of service information of the link, when looking from the first node to the second node, from the first C-bit keep alive message.

3. The method according to Claim 1 , further comprising: constructing a second C-bit keep alive message for a the second node of the neighboring node pair; embedding the second C-bit keep alive message within the C-bit of a second DS3 signal; determining a quality of service information for the link when looking from the second node to the first node; embedding the quality of service information within the C-bit keep alive message; and transmitting the second DS3 signal from the second node to the first node in the neighboring node pair over the link to identify the second node to the first node and the quality of service information for the link, when looking from the second node to the first node, from the second C-bit keep alive message.

4. The method of Claim 3, wherein the first node has a first identification designation and the second node has a second identification designation and the link quality of service information, and further wherein the first C-bit keep alive message contains the first identification designation and the second C-bit keep alive message contains the second identification designation and the link quality of service information.

5. The method according to Claim 4, wherein the first identification designation includes a first node identifier, a first node port number, a first node a wide area network address, and a first node "I am custodial" indicator in the first C-bit keep alive message, and wherein the second identification designation includes a second node identifier, a second node port number, a second node a wide area network address, and a second node "I am custodial" indicator in the second C-bit keep alive message.

6. The method of Claim 1 , wherein the network comprises a plurality of neighboring node pairs, and further wherein a C-bit embedded keep alive message is sent between each pair of neighboring nodes connected by a link.

7. The method of Claim 1 , wherein each node has an identification designation, and further wherein each C-bit embedded keep alive message includes the identification designation for the node from which the C-bit embedded keep alive message originates and the quality of service information of the link looking from one node to the next node.

8. The method according to Claim 7, wherein the link is a spare link.

9. The method according to Claim 3, wherein the link is a spare link.

10. The method according to Claim 2, wherein the link is a spare link.

11. The method according to Claim 1 , wherein the link is a spare link.

12. The method of Claim 7, wherein each identification designation further comprises a node identifier, a port number, a wide area network address, and an "I am custodial" indicator for the node from which the C-bit embedded keep alive message originates.

13. The method of Claim 12, wherein each DS3 signal travels in-band.

14. The method according to Claim 12, wherein the quality of service information of the link includes at least one of errored seconds, severely errored seconds, and loss of signal seconds.

15. The method of Claim 8, wherein each identification designation further comprises a node identifier, a port number, a wide area network address, and an "I am custodial" indicator for the node from which the C-bit embedded keep alive message originates.

16. The method of Claim 15, wherein each DS3 signal travels in-band.

17. The method according to Claim 16, wherein the quality of service information of the link includes at least one of errored seconds, severely errored seconds, and loss of signal seconds.

18. The method according to Claim 17, wherein the errored seconds, severely errored seconds and loss of signal seconds and quality of service are defined by the DS3 standard.

19. A telecommunications network comprising a plurality of nodes interconnected by a plurality of links, and a distributed restoration sub-network, comprising: a first node having a first unique identifier; a second node having a second unique identifier; a link connecting the first node to the second node; and a DS3 signaling channel within the link, wherein the first node and second node are operable to send a DS3 signal having a keep alive message and a quality of service information of the link embedded within a C-bit to one another.

20. The network of Claim 19, wherein a DS3 signal from the first node contains the first unique identifier and a DS3 signal from the second node contains the second unique identifier.

21. The network of Claim 20, wherein the first node is operable to send a first node DS3 signal to the second node over the link to identify the first node to the second node, and wherein the second node is operable to send a second node DS3 signal to the first node over the link to identify the second node to the first node.

22. The network of Claim 19, wherein the network further comprises a plurality of neighboring node pairs, and further wherein each pair connected by a spare link is operable to send a plurality of DS3 signals with a C-bit embedded keep alive messages to each other.

23. The network of Claim 22, wherein each node has a unique numerical identifier, and further wherein each keep alive message comprises the unique numerical identifier for the node from which the DS3 signal is generated.

24. The network of Claim 23, wherein each unique numerical identifier includes at least one of a node identifier, a port number, a wide area network address, and an "I am custodial" indicator for the node from which the DS3 signal is generated.

25. The network of Claim 19, wherein each node has a unique numerical identifier, and further wherein each keep alive message comprises the numerical identifier for the node from the keep alive message originators.

26. The network of Claim 19, wherein the keep alive message includes at least one of a node identifier, a port number, a wide area network address, and an "I am custodial" indicator.

27. The network of Claim 19, wherein all DS3 signals travel in-band.

28. The network according to Claim 27, wherein the errored seconds, severely errored seconds and loss of signal seconds and quality of service are defined by the DS3 standard.