traffic grooming in wdm networks dynamic traffic grooming in wdm mesh networks using a novel graph...

Traffic grooming in WDM Traffic grooming in WDM NetworksNetworks

•Dynamic Traffic Grooming in WDM Mesh Networks Using a Novel Graph Model by

Hongyue Zhu, Hui Zang, Keyao Zhu, and Biswanath Mukherjee

What is Traffic Grooming?What is Traffic Grooming?

• When low speed traffic streams are When low speed traffic streams are multiplexed and switched onto high-multiplexed and switched onto high-speed light paths, we say traffic is speed light paths, we say traffic is groomed.groomed.

• Grooming is mainly done to reduce Grooming is mainly done to reduce the no of Add Drop Multiplexers the no of Add Drop Multiplexers (ADM) required. As they are major (ADM) required. As they are major contributors to the total cost.contributors to the total cost.

Motivation for Traffic Motivation for Traffic GroomingGrooming•Suppose that each wavelength is used

to support anOC-48 ring, and that the traffic requirement is for eight OC-3 circuits between each pair of nodes. In this example we have six node pairs, and the total traffic load is equal to 48 OC-3s or equivalently three OC-48 rings. In the next slide 2 possible assignments are shown.

Motivation for Traffic GroomingMotivation for Traffic Grooming(cond..)(cond..)

• Thus we can see that by careful Thus we can see that by careful selection of wavelengths passing selection of wavelengths passing through a node we can reduce the through a node we can reduce the required no of ADM’S.required no of ADM’S.

• Application of RAW alone does not Application of RAW alone does not imply that the solution selected is imply that the solution selected is optimal in no of ADM’S required. optimal in no of ADM’S required. Consider the example on the next slide.Consider the example on the next slide.

• As shown RAW 1 requires only 2 As shown RAW 1 requires only 2 wavelengths and 15 AMD’S. While RAW wavelengths and 15 AMD’S. While RAW 2 requires 3 wavelengths , but 2 requires 3 wavelengths , but consume only 9 AMD’S.consume only 9 AMD’S.

• Generally a traffic grooming problem Generally a traffic grooming problem can be formulated as an ILP. But as the can be formulated as an ILP. But as the network size grows the no of equations network size grows the no of equations and variables increase explosively.and variables increase explosively.

Novel Graph ModelNovel Graph Model

• Auxiliary graph: Captures various network Auxiliary graph: Captures various network constrains, like no of transceivers at each constrains, like no of transceivers at each node, no of wavelengths on each fiber-link, node, no of wavelengths on each fiber-link, wavelength-conversion capabilities of each wavelength-conversion capabilities of each node etc (will be discussed in details) node etc (will be discussed in details)

• Dynamic traffic grooming Algorithm: This Dynamic traffic grooming Algorithm: This is a route computing algorithm, which take is a route computing algorithm, which take the weight function into account. Thus by the weight function into account. Thus by dynamically adjusting the weight’s on the dynamically adjusting the weight’s on the edges, one could evolve from one edges, one could evolve from one grooming policy to another, as demand grooming policy to another, as demand changes. changes.

Construction of an Auxiliary Construction of an Auxiliary GraphGraph

• Consider a network of 3 nodesConsider a network of 3 nodes• Each link has two wavelengthsEach link has two wavelengths• All nodes are assumed to have grooming All nodes are assumed to have grooming

functionalityfunctionality• Node 0 has full wavelength-conversionNode 0 has full wavelength-conversion• Node 1 has no wavelength-conversionNode 1 has no wavelength-conversion• Node 2 has limited wavelength-conversion Node 2 has limited wavelength-conversion

capability (i.e. wavelength 1 can be capability (i.e. wavelength 1 can be converted to wavelength 2)converted to wavelength 2)

Construction of an Auxiliary Construction of an Auxiliary GraphGraph(cond..)(cond..)• Auxiliary graph is a layered graph Auxiliary graph is a layered graph

with w+2 layers, where w = no of with w+2 layers, where w = no of wavelengthswavelengths

• W+1 layer is called the W+1 layer is called the Light path Light path LayerLayer

• W+2 layer is called the W+2 layer is called the Access layer.Access layer.

• Each node layer has 2 vertices input Each node layer has 2 vertices input (I) and output (o). (I) and output (o).

Construction of an Auxiliary Construction of an Auxiliary GraphGraph(cond..) Meaning of Edges(cond..) Meaning of Edges• Wavelength Bypass Edges (WBE). There is an edge

from the input port to the output port on each wavelength layer l at node i, denoted as WBE (i, l).

• Grooming Edges (GrmE). There is an edge from the input port to the output port on access layer at node I if node i has grooming capability, denoted as GrmE (i).

• Mux Edges (MuxE). There is an edge from the output port on the access layer to the output port on the lightpath layer at each node, denoted as MuxE(i).

Construction of an Auxiliary Construction of an Auxiliary GraphGraph(cond..) Meaning of Edges(cond..) Meaning of Edges• Demux Edges (DmxE). There is an edge

from the input port on the lightpath layer to the input port on the access layer at each node, denoted as DmxE (i).

• Transmitter Edges (TxE). There is an edge from the output port on the access layer to the output port on wavelength layer l, denoted as TxE(i, l), if there are transmitters available on wavelength λi at node i.

Construction of an Auxiliary Construction of an Auxiliary GraphGraph(cond..) Meaning of Edges(cond..) Meaning of Edges• Receiver Edges (RxE). There is an edge from

the input port on wavelength layer l to the input port on the access layer, denoted as RxE(i, l), if there are receivers available on wavelength λi at node i.

• Converter Edges (CvtE). There is an edge from the input port on wavelength layer l1 to the output port on wavelength layer l2 at node i, denoted as CvtE(i, l1, l2), if wavelength l1 can be converted to wavelength l2 at node i.

Construction of an Auxiliary Construction of an Auxiliary GraphGraph(cond..) Meaning of Edges(cond..) Meaning of Edges• Wavelength-Link Edges (WLE). There is an

edge from the output port on wavelength layer l at node i to the input port on wavelength layer l at node j, denoted as WLE(i, j, l), if there is a physical link from node i to node j and wavelength λl on this link is not used.

• Lightpath Edges (LPE). There is an edge from the output port on the lightpath layer at node i to the input port on the lightpath layer at node j, denoted as LPE(i, j), if there is a lightpath from node i to node j.

Construction of an Auxiliary Construction of an Auxiliary GraphGraph(cond..)(cond..)

– Each edge is associated with the tuple P(c,w).Each edge is associated with the tuple P(c,w).– For wavelength-link edges c = capacity of the For wavelength-link edges c = capacity of the

corresponding wavelength on the corresponding wavelength on the corresponding link.corresponding link.

– For lightpath edges c = residual capacity of For lightpath edges c = residual capacity of corresponding lightpath.corresponding lightpath.

– For all other type of edges c = infinity.For all other type of edges c = infinity.– Weight w reflect cost of element.Weight w reflect cost of element.– Weights can be fixed of adjusted in accordance Weights can be fixed of adjusted in accordance

to network stateto network state– Fixed weight Fixed weight Fixed grooming policy Fixed grooming policy– Variable weight Variable weight Adaptive grooming policy Adaptive grooming policy

Auxiliary GraphAuxiliary Graph

Dynamic traffic grooming Dynamic traffic grooming AlgorithmAlgorithm

• Inputs:Inputs:

1.1. Initial network stateInitial network state

2.2. Set of traffic demands represented Set of traffic demands represented as as

T (s, d, g, m). s = source, d= T (s, d, g, m). s = source, d= destination, g= granularity of traffic destination, g= granularity of traffic and m = amount of traffic in g units.and m = amount of traffic in g units.

Algorithm stepsAlgorithm steps

• Initialize: Construct auxiliary graph.Initialize: Construct auxiliary graph.• When request T arrivesWhen request T arrives1 1 Compute the shortest path p from the output

port on the access layer of the source to the input port on the access layer of the destination of T on graph G, ignoring the edges whose capacities are less than the requirement of the request. If such a path does not exist, block the traffic demand; otherwise, continue with the following steps.


2 2 If p contains wavelength-link edges, set up one or more lightpaths going through the corresponding wavelength-links.

3 Route T along the pre-existing lightpaths in p and/or lightpaths newly set up according to p.

4 Update graph G as follows:


• For each newly setup lightpath, a lightpath edge from the output port of the starting node of the lightpath to the input port of the ending node of the lightpath is added on the lightpath layer.

• The wavelength-link edges used by the lightpath are removed from the corresponding wavelength layers.


• If there is no more transmitter/receiver available at node i on wavelength λl , the corresponding transmitter/receiver edge will be removed from G, i.e., this node cannot source/sink a lightpath on wavelength λl any more and can only be bypassed by a lightpath.

• If there is no more wavelength converter which can convert wavelength l1 to wavelength l2 available at node i, the converter edge will be removed from G.

• Update tuple P(c,w)


5 If connection removed5 If connection removed

A Remove the traffic from network.A Remove the traffic from network.

B Tear down all the lightpathsB Tear down all the lightpaths

C Update graph G by applying reverse C Update graph G by applying reverse of update methods used in step 4 of update methods used in step 4 above.above.

ExampleExample

• Assume:Assume:• Capacity of each wavelength = OC-48Capacity of each wavelength = OC-48• Each node has grooming capability and Each node has grooming capability and

two tunable transceivers.two tunable transceivers.• First connection request = First connection request = T(1, 0, OC-12, 2)T(1, 0, OC-12, 2)Path found :Path found :TXE(1,1) WLE(1,0,1) and RXE(0,1)TXE(1,1) WLE(1,0,1) and RXE(0,1)LPE(0,1) = 24LPE(0,1) = 24

ExampleExample

ExampleExample

• Another request:Another request:T(2,0,OC-12,1)T(2,0,OC-12,1)Path found:Path found:• Case1: Case1: TxE(2,2), WLE(2,1,2), WBE(1,2),

WLE(1,0,2), and RxE(0,2)LPE(2,0) = 36 LPE(1,0) = 24• Case2: TxE(2,1), WLE(2,1,1), RxE(1,1),

GrmE(1), MuxE(1), LPE(1,0), and DmxE(0)LPE(2,1) = 36 AND LPE(1,0) = 12

case1case1

Case 2Case 2

Grooming OperationsGrooming Operations

• Op1:Op1:Route the traffic onto an existing lightpath directly connecting the source s and the destination d.

• Op2: Route the traffic through multiple existing lightpaths.

• Op3: Set up a new lightpath directly between the source s and the destination d and route the traffic onto this lightpath. Using this operation, we set up only one lightpath if the amount of the traffic is less than or equal to the capacity of the lightpath.

Grooming OperationsGrooming Operations

• Op4: Op4: Set up one or more lightpaths that do not directly connect source s and destination d, and route the traffic onto these lightpaths and/or some existing lightpaths. Using this operation, we need to set up at least one new lightpath. However, since some existing lightpaths may be utilized, the number of wavelength-links used to set up the new lightpaths could be less than the number of wavelength-links needed to set up a lightpath directly connecting source s and destination d.

Grooming PoliciesGrooming Policies

• By combining various grooming operations in By combining various grooming operations in different priority order , we can achieve different priority order , we can achieve different grooming policiesdifferent grooming policies

1. Minimize the Number of Traffic Hops on the Virtual Topology (MinTHV) : This policy chooses the route with the fewest lightpaths for a connection.

2. Minimize the Number of Traffic Hops on the Physical Topology (MinTHP) : We compare the number of wavelength-links used by all the four operations and choose the one with the fewest wavelength-links.

Grooming PoliciesGrooming Policies

3. 3. Minimize the Number of Lightpaths (MinLP) : This policy is similar to MinTHV but it tries to set up the minimal number of new lightpaths to carry the traffic.

4. Minimize the Number of Wavelength-Links (MinWL) : This policy is similar to MinTHP but it tries to consume the minimum number of extra wavelength-links, i.e., wavelength-links not being used by any lightpaths for now, to carry the traffic

Dominant edgeDominant edge

• if a path p1 in the graph contains more of this kind of edges than another path p2, then the weight of p1 is always larger than that of p2. Here, the weight of a path is the summation of the weights of the edges it traverses.

• Example: • To achieve MinTHV, we just need to make GrmEs

the dominant edges.• To achieve MinLP, we should make TxEs and RxEs

the dominant edges.• To achieve MinWL, WLEs should be the dominant

edges

ResultsResults

Adaptive grooming policyAdaptive grooming policy

•Since MinTHV performs best when transceivers are the more constrained resources and MinTHP gives the best results when wavelength-links become more scarce resources, Adaptive Grooming Policy (AGP) take advantages of both these policies and performs well over all network conditions.


• ratio of the number of unused wavelength-links in the network to the total number of available transceivers at all nodes as an indicator of the network state. If the ratio is larger than the set threshold d1 then MinTHV will be used and if the ratio is less that the set threshold d2 then MinTHP will be used. If ratio is in between then the policy is not changed.

ConclusionConclusion

• The new model takes various The new model takes various constrains into account and can constrains into account and can achieve various objectives by using achieve various objectives by using different grooming policies. The different grooming policies. The ability to adjust grooming policy by ability to adjust grooming policy by changing the weights of the edges changing the weights of the edges makes this model very suitable for makes this model very suitable for dynamic traffic grooming.dynamic traffic grooming.

traffic grooming in wdm networks dynamic traffic grooming in wdm mesh networks using a novel graph...

Documents