Existing networks use WDM as a physical media layer for point-to-point transmission. A lightpath is an optical communication path between two nodes, . A viable solution to meet this challenge is to use optical networks based on wavelength-division multiplexing (WDM) technology. WDM divides the vast. Lightpath Restoration in WDM Optical Networks A Survey in IEEE Network Magazine Nov/Dec Wavelength Division Multiplexing (WDM) • Why? To meet.
|Published (Last):||19 March 2010|
|PDF File Size:||12.90 Mb|
|ePub File Size:||14.69 Mb|
|Price:||Free* [*Free Regsitration Required]|
Skip to main content. Log In Sign Up. Active light-path restoration in WDM networks [Invited. Azim, Xiaohong Jiang, Md. As these networks carry gigantic amounts of rextoration, survivability has become a primary and basic need for them. We propose a novel active restoration scheme that restores a failed primary lightpath based on multiple predefined but nonreserved backup paths along the primary path.
Extensive simulation results based wddm three typical test networks indicate that with a small increase in restoration time, our new restoration scheme can reduce the blocking probability and the capacity requirement significantly while guaranteeing a very high restoration probability.
Introduction Wavelength-division multiplexing WDM networks have the capability of providing huge bandwidth, and it networos expected that WDM will be a dominant technology for the next- generation optical Internet. As WDM networks carry more and more data, failure of any part and the resulting inability to move data around quickly may have a tremendous eco- nomic effect.
Todimala and Ramamurthy  pointed out that fiber cuts occur at the rate of 4. For this reason, survivability issues in high-bandwidth WDM networks have become an important area of research in recent years. Survivability is defined as the ability of the network to maintain service continuity to end users in the presence of network failures .
The approaches to ensuring survivability can be generally classified as proactive protection and reactive restoration. With the former, a backup lightpath is computed, and wavelength channels are reserved for it at the restoratoon the primary lightpath is established. If both primary and backup lightpaths are available for a demand, the demand is accepted. In reactive restoration optixal backup lightpath is searched libhtpath the primary lightpath is interrupted.
Nftworks lightpath restoration schemes for WDM optical networks lighfpath reported recently [7, 8]. Although reactive restoration is more efficient in terms of capacity use and blocking probability, it may lead to an unacceptably long restoration time because a global search for a backup lightpath is required.
As proactive protection experiences a high blocking probability and huge network re- source redundancy, whereas the reactive restoration dwm in a unsatisfactory long restora- tion time, the motivation of this study is to find a compromise between the proactive protec- tion and reactive restoration schemes so that good performance can be achieved. In partic- ular, we propose a novel active restoration scheme in which a primary lightpath is guarded by multiple backup lightpaths that are predefined but not reserved along the primary path before failure occurs.
The selection of the backup path to be employed in the restoration process relies on the failure location as well as the availability of resources.
The related routing and wavelength-assignment algorithms for the proposed scheme are also developed in this paper. Simulation results based on three well-known networks show that our new scheme, compared with the proactive protection scheme, can reduce the blocking probabil- ity and save network resources significantly, with only small sacrifices in restoration time and restkration probability.
The rest of the article is organized as follows: Section 2 presents the problem prelim- inaries of this work. Section 3 deals with our new scheme and the related routing and wavelength-assignment algorithms. Section 4 presents the simulation results and discus- sions, and Section 5 presents the concluding remarks.
We assume that there is at most one bidirectional fiber between any two nodes and each fiber carries the same number of wavelengths. The adjacency matrix A of G can be represented as follows: The set of paths Ts,d between a node pair vs and vd can be represented as a sequence of distinct nodes as follows: If we use E P to denote the set of edges a path P travels, then the length L Wvm of the path P can be computed as follows: The length of a primary path Pp between a node pair vs and vd can be represented as follows: The destination node will conduct a destination-initiated resource reservation mechanism, notifying the source node to set up the backup lightpath so that the affected opticl traffic can be restored.
Figure 1 exemplifies this case. Let us consider v1v2v3v4v5 the primary path. If the link between v1 and v2 is cut, v2 will be subject to a LOL failure and send a FNM to the destination node along the primary path. The destination node v5 then starts to set up the backup lightpath.
Active Light-Path Restoration Scheme In this section we first introduce our novel active restoration scheme, and then we estimate its restoration time. The complete routing and wavelength-assignment algorithm for our scheme is also included in this section. Active Light-Path Restoration Iptical consider the case in which traffic demands or requests arrive one by one without any prior knowledge of future arrival. In our new scheme we compute for each connection request a primary lightpath.
If enough wavelength channels are available along the primary path, we then compute multiple backup paths that start from the nodes along the primary path and end at the source node of the path, respectively as we show in Otical 3. Here a backup path starting from a node of the primary path is just the shortest path from that node to the source node that is link-disjointed with the primary path.
In this way our proposed scheme guarantees that the selected backup path will never retain any faulty link of the failed primary path.
Lightpath Restoration in WDM Optical Networks
Networkks the other hand, if the primary path cannot be established because of a shortage of resources, the connection is blocked. When there is a link failure along the primary path, the immediate down- stream node next to the failure checks the availability of the successive backup paths and may send a FNM to the corresponding supported node along the primary path.
If the backup path starting from the first node next to the failure is available, then this node will only send a setup message directly to the source node.
Otherwise, the FNM may be forwarded to fol- lowing supported nodes up to the last node.
We refer to the supported node through which traffic restoration can be performed as a restoration node. Figure 2 is an example that clar- ifies the main idea of our scheme. Let node v1 be the source node, node v5 be the destination node, and the primary path be v1v2v3v4v5. If the link between v1 and v2 fails, the node v2 will detect a LOL failure. Since node v2 does not have any backup path to the source node that is link disjointed with the primary path, node v2 is an unsupported node.
According to the availability of the wavelength channels along the successive backup paths, node v2 will decide which backup path will be employed in the restoration of the disrupted traffic assuming that every node keeps all information about other nodes in the network.
If wavelength channels are available along the backup path [e. As soon as node v3 receives the FNM, it will immediately send a setup message to the source node through the backup path v3v6v1. Once the source node v1 receives the setup message, it reroutes all data to the backup path v1v6v3then data will go through the rest of primary path to the destination node. If there are not enough resources along the backup path of the first supported node, the backup paths of the following supported nodes will be investigated [e.
If no backup path is available because of a lack of network resources, restoration of this lightpath fails. Restoration Time The restoration process of our scheme consists of two phases.
Path Determination and Failure Notification For a given primary path, the first phase of our scheme involves finding the first available backup path among the set of the precomputed backup paths.
The time cost of this phase is related to the computational power of network nodes. Backup Path Activation The second phase of our scheme is to activate the reserved backup path. Let Lsetup be the length that the setup message needs to travel and Pb be the backup path initiated from the first restoration node.
The restoration time is defined as the elapsed time from the instant a connection is disrupted to the instant it is restored . The restoration time RT of our scheme, which is proportional to the sum of the FNM length and Lsetupcan be represented as follows: Routing and Wavelength Assignment The RWA algorithm of our scheme involves calculating the primary light path and its re- lated multiple backup paths in advance of failure.
The corresponding multiple backup paths are derived by erasing the links of the primary path from the network topology and then performing the Dijkstra algorithm on the reduced network topology again. Therefore, the computation complexity of finding the multi- ple backup paths for our proposed scheme is identical to that of the proactive path-based algorithm.
The routing algorithm of our scheme consists of two phases as follows: Select the immediate downstream node of the failed link along the primary path as CN. If CN is supported. IF WA algorithm succeeds. Then an algorithm is executed, in which, CN, NN, and DN denote the current node being checked, the immediate downstream node of CN, and the destination node along the primary path, respectively.
Wavelength Assignment The task of wavelength-assignment algorithm is to assign a wavelength to each link along the backup route. In our approach, once a failure occurs in the primary path, the wavelength-assignment algorithm will try to assign wavelength channels to all links of the backup path initiated from the first supported node.
If the wavelength-assignment algo- rithm succeeds in reserving the required resources along the backup path, the disrupted traffic will be restored from the first supported node and the after-failure wavelength as- signment process is stopped. Otherwise, the wavelength-assignment process will continue for the successive supported nodes until the last one. If the wavelength-assignment algo- rithm fails to reserve any of these backup paths along the primary path, the restoration for the disrupted primary path fails.
Under the assumption that each network node has the capability of full wavelength conversion, the after-failure wavelength assignment is given by Algorithm 2. RWA Time Complexity In our active lightpath restoration scheme, the before-failure routing phase is responsible for computing the primary path and the backup route segments from the source to every other node along the primary path.
The after-failure routing phase in our scheme is responsible for selecting the first available backup path. The worst case time complexity of this phase is O Vwhich corresponds to the case when the failure occurs at the immediate downstream link of the source node and the primary path has a hop count of V.
FOR the backup path Bi 3.
Lightpath Restoration in WDM Optical Networks
FOR each link l on Bi 6. Restore the traffic Assign w to l and GOTO 6 Terminate a very high probability by inspecting only the first three predefined backup paths. Performance Evaluation In this section we present extensive simulation results to validate our new scheme.
The workload measured in Erlangs is the product of the average arrival rate and the mean holding time of requests. We assume that all nodes have the full- wavelength conversion capability and each fiber has 32 wavelengths. A request is equally likely to have any pair of network nodes as its source destination.
A total of five batches was simulated, with sessions generated in each batch.
Table 1 summarizes the characteristics of the test networks, including the number of nodes, the number of links, and the average nodal degree defined as the number of links terminating at the node . For iptical comparison, we use three performance met- rics of blocking probability, restoration probability, and restoration time.
Nodal Degree 1 14 21 3 2 32 51 3.
Simulation Results We first show in Figs. The results shown in Figs. When the workload is 7.