Protection Mechanisms for Optical WDM Networks based on

  • Slides: 38
Download presentation
Protection Mechanisms for Optical WDM Networks based on Wavelength Converter Multiplexing and Backup Path

Protection Mechanisms for Optical WDM Networks based on Wavelength Converter Multiplexing and Backup Path Relocation Techniques Sunil Gowda and Krishna M. Sivalingam University of Maryland Baltimore Country(UMBC) Dept. of CSEE, Baltimore Presented by: Priyanka Das 1

Focus n This paper studies the problem of designing a survivable optical WDM network.

Focus n This paper studies the problem of designing a survivable optical WDM network. n Focus here is on efficient use of optical converters. n n Mechanism for improving network performance for survivable WDM mesh networks. An enhancement of dynamic route computation mechanism. Goal To minimize the number of converters per node used in the optical WDM network 2

Primary and Backup Route Computation Mechanisms n Conversion free primary routing (CFPR). n Converter

Primary and Backup Route Computation Mechanisms n Conversion free primary routing (CFPR). n Converter multiplexing. n Backup path relocation. All these mechanisms attempt to improve the overall performance of the network. 3

Conversion Free Primary Routing (CFPR) n This routing scheme is proposed to compute wavelength

Conversion Free Primary Routing (CFPR) n This routing scheme is proposed to compute wavelength conversion free primary paths as far as possible. The basic objective here is to: : n Reduce the number of converters used in the network. n Reduce cost. n Eliminate conversion delay. n Avoid signal degradations. 4

Converter Multiplexing: “ Technique which allows wavelength converters to be shared among multiple backup

Converter Multiplexing: “ Technique which allows wavelength converters to be shared among multiple backup paths”. Objective n To reduces the number of connections blocked due to the unavailability of wavelength converters and reduces numbers of converters in use, thereby limiting the expenditure. 5

Backup Path Relocation (BPR) BPR is used when it becomes necessary for primary paths

Backup Path Relocation (BPR) BPR is used when it becomes necessary for primary paths to accommodate certain routes which are occupied by the backup path, at such a situation the backup paths are migrated so some other wavelength or segment. Objective n This helps in providing primary paths with fewer hops. n Reduces blocking. n Improves network utilization. 6

Improving Network Performance n Reducing the number of converters used per node in the

Improving Network Performance n Reducing the number of converters used per node in the network. n Making it cost effective. n Providing protection against failure of primary path. n Reducing blockage in the network with the help of shared converters. 7

Routing and Wavelength Assignment Problem n Here dynamic network model is used where requests

Routing and Wavelength Assignment Problem n Here dynamic network model is used where requests arrive dynamically. n Each request specifies source, destination and bandwidth required. n Each request is then assigned a lightpath for a path/wavelength combination for it’s entire duration. The problem of determining end-end route and wavelength is referred to as RWA problem . 8

Wavelength Router Architecture Wavelength constraint is removed by using wavelength converters. Lightpath can thereby

Wavelength Router Architecture Wavelength constraint is removed by using wavelength converters. Lightpath can thereby use different wavelengths on different links of the path. But converters are n Expensive n Produce signal degradation and delay So here the focus is on Minimizing the usage of wavelength conversion for primary path and thus reduce the number of converters used. 9

Wavelength Converter Switch Architecture n There are three different architectures proposed for a wavelength

Wavelength Converter Switch Architecture n There are three different architectures proposed for a wavelength convertible switch. n Dedicated wavelength converter switch architecture. n Share -per-node architecture. n Share-per-link architecture. n n The performance of share-per-node is better than dedicated in terms of cost and high utilization but it is complex due to higher switching complexity and blocking due to unavailability of converters. The performance of share-per-link in terms of cost lies in between the other two. 10

Share-Per-Node Architecture n n WBC is the wavelength converter bank and it is provided

Share-Per-Node Architecture n n WBC is the wavelength converter bank and it is provided for the entire router. It provides best cost to performance ratio 11

Protection in Mesh Topology in WDM Networks Types of failures : n Link failure:

Protection in Mesh Topology in WDM Networks Types of failures : n Link failure: This needs rerouting of lightpath on the affected link n Node failure: The affected lightpath is handled by other nodes. 12

Protection vs Restoration n n Works in advance Lower recovery time Needs redundant spare

Protection vs Restoration n n Works in advance Lower recovery time Needs redundant spare capacity Offers guarantee Also called as PROACTIVE n n n Functions after failure More recovery time More resource utilization Cannot offer 100% guarantee. Also called as REACTIVE recovery protection Link level restoration Path level Link level Path level 13

Protection protection Dedicated protection scheme 1: 1 protection Shared protection scheme 1: M protection

Protection protection Dedicated protection scheme 1: 1 protection Shared protection scheme 1: M protection Dedicated path for each individual connection Wavelength can be shared among multiple backup paths 14

Backup Path Multiplexing n n 1: M protection, i. e. one wavelenght can be

Backup Path Multiplexing n n 1: M protection, i. e. one wavelenght can be shared by many backup paths provided they are both never activated simultaneously. This provides 100% restoration guarantee in case of single ling failure. 15

Lightpath Migration n Migration of lightpath onto new paths, to accommodate other connections is

Lightpath Migration n Migration of lightpath onto new paths, to accommodate other connections is the basic concept used. A virtual topology reconfiguration scheme to adapt to the changing traffic pattern s has been modeled as an Integrated linear programming (ILP) formulation n Light paths are however torn down and re-established on the new paths. n During the reconfiguration the transmission on that path is terminated. n This helps to provide better paths for the primary. n 16

Network Architecture n n Dynamic routing is used, where the shortest path is computed

Network Architecture n n Dynamic routing is used, where the shortest path is computed between nodes based on current situation. Path level protection is used with both dedicated and shared protection schemes for backup paths. Wavelength route architecture is based on share-per-node wavelength converter configuration, as it offers best cost to performance ratio. Connections are blocked only due to unavailability of free wavelength or wavelength converters. 17

How does things happen? n Step 1: Request arrives with all the specification. n

How does things happen? n Step 1: Request arrives with all the specification. n Step 2: Conversion free primary routing. n Step 3: If step 2 is not possible then use hop-count based shortest path algorithm. n Step 4: Working on step 3 needs wavelength conversions and hence blocking due to less number of converters available. n Step 5: Here converter multiplexing is proposed. n Step 6: Backup path relocation comes to picture when needed. 18

Conversion Free Primary Routing (CFPR) Technique. n Aim: To avoid wavelength conversions while routing

Conversion Free Primary Routing (CFPR) Technique. n Aim: To avoid wavelength conversions while routing primary connections. n Multi-layered graph is used, these layers represent individual wavelength planes. n n CFPR algorithm models such a graph although the network has conversions capabilities. For each wavelength plane, the nodes are the physical nodes. 19

Notations Existence of edges if wavelength is either not allocated or is reserved for

Notations Existence of edges if wavelength is either not allocated or is reserved for some backup path(s). if wavelength w on link (i, j) is assigned to primary Computation of routes is done by Dijkstra’s shortest path. denotes the shortest path from node s to node d wavelength w. if no path is available on this wavelength The routing scheme here calculates up to W paths, one on each wavelength. 20

Conversion Free Primary Routing and Overlapping n 21

Conversion Free Primary Routing and Overlapping n 21

Advantages of CFPR Mechanism n Reduces conversion delays and degradation due to converters. n

Advantages of CFPR Mechanism n Reduces conversion delays and degradation due to converters. n Lower computational complexibility. n CFPR computes path on each wavelength separately and hence alternate paths are available if shorted paths are blocked 22

Converter Multiplexing n n Based on backup path multiplexing. Converters are shared only among

Converter Multiplexing n n Based on backup path multiplexing. Converters are shared only among backup paths that have physically disjoint primary paths. The converters are reserved during the establishment of the backup paths and are tuned to required wavelength during recovery. Source sends CONV-RESV message to the node at which conversion is needed. 23

Converter Multiplexing n The node responds with CONV-RESV-ACKS if accepted. n The node responds

Converter Multiplexing n The node responds with CONV-RESV-ACKS if accepted. n The node responds with CONV-RESV-NACK if not accepted. n Backup path is completed by the source node only if it receives all such acknowledges. n A wavelength conversion status table (WCST) is maintained at each node. n When network fails, for path recovery initialization CONV-SETUP message is sent to the node to configure the converter. 24

Example of Converter Multiplexing n n Path p 1(1 -6 -7 -8) and p

Example of Converter Multiplexing n n Path p 1(1 -6 -7 -8) and p 2 (4 -8) are the primary paths. The corresponding backup paths are b 1(1 -2 -58) and b 2(4 -5 -8). Since the primary paths are link disjoint, the backup paths can share a wavelength converter at node 5. Due do converter multiplexing the number of converters is reduced from 2 to 1 at node 5. 25

Backup Path Relocation Two relocation schemes are proposed to migrate an overlapping backup segment.

Backup Path Relocation Two relocation schemes are proposed to migrate an overlapping backup segment. n n The wavelength relocation (WR) : New wavelength is used for the overlapping segment. The segment relocation (SR): Overlapping segment is relocated on a completely different path. 26

Understanding the difference between WR and SR. 27

Understanding the difference between WR and SR. 27

Wavelength Relocation vs. Segment Relocation n n n WR is simpler since the overlap

Wavelength Relocation vs. Segment Relocation n n n WR is simpler since the overlap segment’s links are unchanged Control messages have to be sent only to the nodes of the overlapping segment about the configuration. However, free wavelengths may always be not available on the same set of links, resulting in relocation failure SR considers a large set of paths and offers higher success probability relocation. However, such relocation incurs large overhead as overlapping segments are released and re-established. And consumes more resources due to potentially longer paths. 28

Performance Analysis Simulation model n n n n n Dynamic network traffic Request arrive

Performance Analysis Simulation model n n n n n Dynamic network traffic Request arrive at the node according to Poisson process with rate λ There is uniform node destination distribution Each request is assignment a wavelength Traffic load is L, and it is defined as λ/ µ in Erlangs. session duration is exponentially distributed with a mean of 1/µ. Share-per-node architecture is used. And C denotes # of converters. Dedicated and shared protection schemes are studied. Single link failure model is assumed. 29

Simulation Models Simulation are performed for two networks: n n A 24 -node ARPANET-like

Simulation Models Simulation are performed for two networks: n n A 24 -node ARPANET-like network with 16 and 32 wavelength on each link. The results for this network is discussed here. A random 50 -node network with 32 wavelength per link. 30

Different Mechanisms n n n The basic hop count (HC) based shortest path routing

Different Mechanisms n n n The basic hop count (HC) based shortest path routing algorithm. The CFPR routing algorithm with wavelength relocation. The CFPR routing algorithm with segment relocation. The notations X-Y-Z is used to specify an algorithm, where X € {HC, CFPR} denotes the routing algorithm Y € {NR, WR, SR} denoted no relocation, wavelength relocation and segment relocation respectively Z € {DP, SP} denotes dedicated and shared protection. The performance metrics presented are the blocking probability (), link and converter utilization, average hop count, and statistics on backup path relocation 31

Blocking Probability. n n Networking blocking probability is defined as the fraction of the

Blocking Probability. n n Networking blocking probability is defined as the fraction of the total connection requests that are rejected. Converter multiplexing and backup path relocation schemes perform better than the basic scheme. Result n CFPR with WR/SR, with dedicated or shared show lower blocking probability than the basic scheme. 32

Reduction in Number of Converters. n n n This graph shows the reduction in

Reduction in Number of Converters. n n n This graph shows the reduction in the number of converters required per node. It is evident that wavelength relocation is better than segment relocation when combined with CFPR and converter multiplexing. Although SR-DP and SR-SP works marginally better but increases the complexity and overhead. 33

Average Hop Count n n n This shows the accepted connections for primary paths.

Average Hop Count n n n This shows the accepted connections for primary paths. In the basic scheme, the primary path exhausts all the converters with increasing load and there is a decrease in the average hop count. Whereas the average hop count with the converter multiplexing based algorithms are steady 34

Revenue metric n n This is based on the number of hops routed. Revenue

Revenue metric n n This is based on the number of hops routed. Revenue metrics is defined as shortest –hop count based on the static topology. Result n The proposed algorithm shows marginal drop in revenue while for the basic scheme the revenue drops when load increases 35

Conversion Statistics n n n One of the primary aim was to provide wavelength

Conversion Statistics n n n One of the primary aim was to provide wavelength conversion free paths for the primary paths. In the basic scheme 30% of the connections need at least one converter. While the proposed algorithm eliminates the need of conversion for the primary paths. 36

Relocation Statistics 37

Relocation Statistics 37

Conclusion n n The CFPR routing algorithm significantly reduced the number of primary connections

Conclusion n n The CFPR routing algorithm significantly reduced the number of primary connections undergoing wavelength conversion. The proposed converter multiplexing scheme reduces the number of connections blocked due to unavailability of wavelength converter. Two different backup path relocation mechanisms were also presented , results show that the combination of the two results in substantial reduction in blocking probability. Lower number of converters were used per node. Between both the relocation schemes , the additional overhead of using segment relocation compared wavelength scheme did not result in much improvement, however segment relocation can be used to allow primary connections to be routed on links offering better transmission quality. 38