Multiplex communications – Fault recovery – Bypass an inoperative channel
Reexamination Certificate
1999-10-12
2003-04-15
Chin, Wellington (Department: 2664)
Multiplex communications
Fault recovery
Bypass an inoperative channel
C370S228000
Reexamination Certificate
active
06549513
ABSTRACT:
BACKGROUND OF THE INVENTION
In today's high-speed communication networks, each cable or fiber carries several thousand voice or data circuits. Such large network capacity provides advantages in terms of lower costs and greater payload flexibility. Fiber optic networks enjoy such advantages and other additional advantages, such as improved transmission quality. Whatever the cable technology—wire or fiber—interruptions in communication service are not uncommon. Networks have been known to suffer damage from backhoes at construction sites, power augers, lightning, rodents, fires, train derailments, bullets, vandalism, car crashes, ship anchors, trawler nets, and diverse other mishaps. Given the ubiquitous nature of communications today with its intimate involvement in business, medicine, finance, education, air traffic control police and other government agencies, and other aspects of modern life, it is imperative that interruptions in network operations be restored as quickly as possible in the event of a failure.
The precarious integrity of networks has been recognized, and approaches have been developed for effecting restoration of cable breaks. One such approach has been the implementation of automatic protection switching (APS). APS systems restore service by switching to a dedicated standby system. That is, there are two complete sets of links installed in an APS system so that each link has a back-up link ready and waiting to serve.
In the context of this specification, the term “link” is intended to indicate any communication path intermediate two adjacent nodes, or communication units, in a communication network. Adjacent nodes are communication units, such as cross connect systems, that are connected by a span. There can be more than one link in a span; a span is the set of all links in parallel between two adjacent nodes.
Another recovery system approach is the self-healing ring. Self-healing rings (SHR) vary in the details of their implementation, but they can be conceptualized as an extension of either a 1:1 (100% redundancy) or a 1:N (greater than 100% redundancy) APS system. APS and SHR systems can effect recovery in 50 to 150 milliseconds. Such rapid recovery is very good, but the cost of such systems is prohibitively high except or the most critical of networks, such certain banking, medical, or stock market systems.
Mesh networks have been recognized a useful in providing flexibility in recovering from network interruptions. A mesh network is a network in which each node may be connected to all other nodes in the network via links to adjacent nodes. By using intelligent internetworking devices, such as nodal multiplexers in a T-carrier network, transmissions may be routed over an alternative path should the primary (direct) path between two sites be interrupted. Such interruption may be occasioned by congestion, or by a physical or electrical failure.
Centralized restoration in a mesh network has been attempted, with the calculation of a restoration path being effected at a central location within the network using data stored at that central location. After determination of the restoration path, the information is promulgated throughout the network for implementation. Such centralized restoration systems have not succeeded in restoring network communications in less than times in excess of one minute. With the high capacity time-sensitive information being carried on networks today, such a recovery time is unacceptably slow.
Distributive restoration in a mesh network is another approach that has been discussed in attempting to accommodate restoration of a network. This distributed approach recognizes that digital cross-connect switches employed at nodes in a mesh network are computers, and they collectively represent considerable processing power embedded in a fabric of multiple communication links. In such a distributed approach, every node (digital cross-connect switch) will perform to effect restoration as required in an apparently isolated manner, with no network-wide knowledge of the system. The independently deduced cross-connection decisions of each node will, in the aggregate, collectively constitute effective multipath rerouting plans.
Most of the distributive restoration systems are less costly than an APS or an SHR system. However, the trade-off is that recovery time is not nearly as fast with the duplicative recovery systems. This stands to reason since there is no dedicated link to which traffic can be routed with very little delay. Most distributive restoration systems depend upon a flooding of the network with messages once an interruption is detected. The flooding messages explore all routes then viable in the network. The route (a series of spans denoted by a concatenation of nodes that establishes a way through the network) is sen according to some predetermined route-choosing criteria. Such criteria may include the first (shortest) path identified, the greatest-capacity path, the inclusion of specified nodes within the path, the greatest path length efficiency, the fastest path, or other parameters.
Distributed restoration systems that determine restoration paths after a failure is detected rarely are capable of effecting restoration in less than one second. Such a delay is still unacceptable.
Grover (W. D. Grover, “Distributed Restoration of the Transport Network”, IEEE Network Management Into the 21
st
Century, Chapter 11, February 1994) proposes distributed preplanning for restoration using a digital restoration algorithm. According to Grover's proposal, a self-healing network protocol is executed for each possible span failure in the network. This is accomplished by a full execution of the self-healing network protocol, but without actually making any cross-connections to effect rerouting. Instead, each node is to record the cross-connections it would have made according to the self-healing network protocol, and save those cross-connections in a table. In such manner, each node will have stored in a table the instructions for that node's portion of the response to the self-healing network protocol for each and every respective span of the network. When a failure occurs, the network promulgates an alerting message and any alerted node having non-null actions in its respective table makes the internal cross-connections between spare ports that are listed in its table.
Grover's proposed distributed preplanning involves storing in a table at each node each and every connection that node must participate in for each and every failure case. Such a table takes a significant amount of computing to amass, and a significant amount of time to complete. Grover himself acknowledges that there is a window of vulnerability on the order of seventeen minutes for a 100-span network. According to Grover, alerting can be accomplished either by an activation loop established through all digital cross-connect system (DCS) nodes, or by disseminating the alert through simple flooding. By either of Grover's alerting schemes full promulgation of that message necessary to effect restoration configuration, as each node “consults” its respective table to determine how to participate in the restoration evolution, takes time and network capacity as well. The complexity of constructing Grover's all-connection tables is also further cause for concern as the more complex an operation is, the more aught with opportunity for error it is. Said another way, as a general rule, the more complex a system, the less robust and reliable it is.
Further, Grover does not address how or when the system updates its information regarding which links in the network are actually spare links and available for use in restoration operations. He provides that links used in a restoration path are identified as “in use”, but no allowance is made to identify when a link is not available for restoration operations for any other reason, such as a system reorientation, new subscribers on the system causing use of an additional (previously unused) link, or similar situ
Chao Tzu-Lien Tim
Darabpour Manouchehr
Alcatel
Chin Wellington
Mondul Donald D.
Schultz William
Sewell V. Lawrence
LandOfFree
Method and apparatus for fast distributed restoration of a... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method and apparatus for fast distributed restoration of a..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and apparatus for fast distributed restoration of a... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3053586