A Survey on Static Routing Wavelength Assignment...

A Survey on Static Routing WavelengthAssignment Considering Physical LayerImpairments in Wavelength Divison MultiplexingOptical Networks

Ganesh Santosh Akula

University of Windsor

Imagine a world of people who are connected through networks either they are far or close to

each other and they could communicate in just “planks time”(that is believed to be the possibleshortest duration) by paying less and obtaining great speed to communicate, this is just what the

researchers are trying to obtain these days. One of the major problems facing optical networksis Routing and Wavelength Assignment (RWA) of optical signals taking into consideration the

impairments caused by the physical layer. A regular RWA which does not consider impairments

is often used to solve the RWA of optical signals but this approach could be improved. Variousresearchers have tried to solve regular RWA under physical impairments in order to improve or

replace it. This survey is about their research.

Categories and Subject Descriptors: A.1.0 [Lightpath Requests Establishment]: Routing andAssigning Wavelength of requests considering Physical Layer Impairments.

General Terms: Routing and Wavelength Assignment, Regenerator Placement, Physical Layer

Impairments.

Additional Key Words and Phrases: Static or Offline, Impairment Aware Routing and Wavelength

Assignment, Physical Layer Impairments, Wavelength Division Multiplexing, Heuristic or ILP

based approach, Optical Networks.

ACM Journal Name, Vol. V, No. N, Month 20YY, Pages 1–0??.

2 · Static RWA Considering Physcial Layer Impairments in WDM Optical Networks.

Contents

1 INTRODUCTION 3

2 APPROACHES 3

2.1 Integer Linear Programming and Heuristic based approaches . . . . 3

2.1.1 Integer Linear Programming and Greedy Heuristic approach 3

2.1.2 Binary Integer Linear Programming and Sequential Heuristicapproach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

2.1.3 Noval Integer Linear Programming and Heuristic approach . 5

2.1.4 Integer Linear Programming and Polynomial Time Heuristicapproach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

2.1.5 Integer Linear Programming and 3-Step,Lightpath Segmen-tation Heuristic approach . . . . . . . . . . . . . . . . . . . . 8

2.1.6 Integer Linear Programming and Heuristic approach . . . . . 10

2.1.7 Mixed Integer Linear Programming and Tabu, Greedy SearchHeuristic approach . . . . . . . . . . . . . . . . . . . . . . . . 11

2.1.8 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

2.2 Integer Linear Programming or Heuristic Based approaches . . . . . 14

2.2.1 Integer Linear Programming approach . . . . . . . . . . . . . 14

2.2.2 Relaxed Integer Linear Programming approach . . . . . . . . 15

2.2.3 Heuristic, Meta-Heuristic or Combinatorial approach . . . . . 16

2.2.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

3 CITED BY 17

4 CONCLUDING COMMENTS 17

5 ACKNOWLEDGEMENT 18

6 ANNOTATIONS 18

6.1 Aparicio-Pardo et al. 2011 . . . . . . . . . . . . . . . . . . . . . . . . 18

6.2 Christodoulopoulos et al. 2009 . . . . . . . . . . . . . . . . . . . . . 20

6.3 Katrinis et al. 2011 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

6.4 Manousakis et al. 2009 . . . . . . . . . . . . . . . . . . . . . . . . . . 23

6.5 Pavon-Marino et al. 2009 . . . . . . . . . . . . . . . . . . . . . . . . 24

6.6 Sengezer et al. 2010 . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

6.7 Sengezer 2012 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

6.8 Sole et al. 2012 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

6.9 Varvarigos et al 2010 . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

6.10 Zhang et al. 2009 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30

7 REFERENCES 31

ACM Journal Name, Vol. V, No. N, Month 20YY.

Ganesh Santosh Akula · 3

1. INTRODUCTION

This survey concerns research which attempts to solve the problem of effectivelyand efficiently routing lightpath signals or requests under physical impairmentswhich cause degradation of signals when travelling in optical medium. If there areefficient ways to solve the problem of routing signals under physical impairmentsthen it would enable more effective transmission of data over optical networks, withreduction in errors during transmission and makes the data more reliable and savecost by reducing the redundant transfer of data.

The published papers were identified using the following resources Google Schoolar,ACM digital library, Microsoft Academic Search and other sources available throughthe leddy library, various keywords were used to search.

Two of the research papers, namely Varvarigos et al. [2009], Zhang et al. [2009]were published in conference proceeding and eleven other papers were published injournals. These are: Christodoulopoulos et al. [2009], Manousakis et al. [2009],Pavon-Marino et al. [2009], Saradhi et al. [2009], Sengezer et al. [2010], Varvagiroset al. [2010], Zhai et al. [2010], Aparicio-Pardo et al. [2011], Katrinis et al. [2011],Sengezer et al. [2012], Sole et al. [2012],

The survey consists of a review of various techniques identified, final commentsand annotations of the ten selected papers. Two main methods were identifiedto solve the question. Lastly there is a concluding part incorporating conclusionstatements and annotations eventually.

Taking into consideration all the ten main papers identified it was discovered thatmost of the authors used both Integer Linear Programming (ILP) and Heuristicapproach or either one of the two techniques that is Integer Linear Programmingbased approach or Heuristic based approach in order to route the lightpath requestsunder physical layer impairments.

2. APPROACHES

As stated earlier two distinct techniques have been to route lightpath requests. Thissurvey focuses on these techniques and are classified into two categories one usesboth ILP and heuristic based approach and other which uses either ILP or heuristicbased approach to solve the problem.

2.1 Integer Linear Programming and Heuristic based approaches

Research papers that are considered in this section use the Integer Linear Pro-gramming (ILP) and the Heuristic based approaches to solve the problem. All ofthis research adopts an ILP and a Heuristic approach to route lightpath requestsunder physical layer impairments. A unique approach was proposed by Zhang etal. [2009] in the early years and is followed by other approaches by various authorswhich are either new, extension or alteration to an existing technique to improvethe efficiency and provide cost effective solution to the problem considered.

2.1.1 Integer Linear Programming and Greedy Heuristic approach. Manousakiset al. [2009] propose this method to minimize blocking of lightpaths with effectiveplacement of regenerators. The authors state that during transmission of an opticalsignal the existence of physical impairments in optical fibers block the connectionrequests from reaching the destination. There should be efficient algorithms to re-



route blocked connections by selective placement as well as selective use of minimumnumber of regenerators which re-shapes, re-amplify and re-time the signal.

The authors refer to previous work of Manousakis et.al [2009] and identify noshortcomings in the approach but point out the shortcomings in general taking theoverall scenario and state that in the previous approaches proposed used regener-ators at every node which do not provide a cost effective solution and is a hugedrawback and few other approaches do not consider the impairment effect at all.

The authors state that a new impairment aware algorithm was designed and im-plemented for routing and wavelength assignment along with solving regeneratorplacement and assignment problem in translucent networks which places regener-ators at selective sites along with the number of regenerators to be placed. Theauthors state that the placement and assigning of regenerators is considered as avirtual topology design problem which utilize various ILP and greedy heuristic al-gorithms to solve it and apart from that the approach also determines sequence ofregenerators to be utilized by the connections having impairments and transformthe non-transparent connection to transparent signal by terminating and regener-ating the signal at the specified 3R intermediate node and minimize the number ofblocked connections.

The authors state that in order to evaluate the performance of the proposed im-pairment aware RWA algorithm numerous simulation experiments were performedwith the IA-RWA algorithm. Virtual topology algorithms were implemented inMatlab and LINDO API was used to solve ILP related problems in order to evalu-ate the feasibility of lightpath. The Q-factor estimator was used in the experiment.The link model has a NRZ-OOK modulation format, 10 Gbps transmission rates,50 GHz channel spacing with span length equals to 80km with Geant-2 networktopology with 34 nodes and 54 bidirectional links.

The authors state that when the blocking ratio is graphed against the numberof available wavelengths and demands the PH-based algorithms shows a blockingprobability ranging from 0.05 to 0.00 and required wavelengths ranging from 90to 110 to reach zero blocking with the number of demands varying from 700 to400. Moreover the PH-ILP sum of regenerator algorithm required 115wavelengthsand 461 regenerators while the VP-ILP sum of regenerator required 135 and 430regenerators to reach zero blocking. In addition, the execution time of the IA-RWAalgorithm for translucent network requires about 3 hours in which application ofIA-RWA algorithm used by transparent network in translucent network consumesmost of the time and the remaining phases requires only few 10s of seconds.

The authors state that with the use of the Q factor, low complexity of the vir-tual topology design problem make the approach efficient for small and large scaleheterogeneous mesh networks. Among various algorithms proposed for the virtualtopology design problem the ILP formulation helps in reducing the maximum num-ber of regenerators to be placed and among the heuristics proposed the physicalHop heuristic exhibit better performance and stable function in all cases.

This paper is cited by Aparicio-Pardo et al. [2011], and Sole et al. [2012].

2.1.2 Binary Integer Linear Programming and Sequential Heuristic approach.Pavon-Marino et al.[2009] propose this technique to minimize blocking of light-paths and improve the efficiency by satisfying more number of lightpath requests.



The authors state that the quality of signal degrades due to physical layer impair-ment and if there are no regenerators there are chances of bit errors to be morewhich in turn weakens the signal strength. The authors state that there should beappropriate techniques which consider impairments during signal transmission andhelp in serving maximum number of connections

The authors do not refer to any previous work.The authors state that an IA-RWA offline algorithmic approach is used which

performs a cross layer optimization between the physical and network layers presentand take into account the directly the dominant impairment effect and upper boundon the interference noise variance it can tolerate is calculated and it is used asconstraint the interfering noise caused by other lightpaths and avoid heavy ILPformulations.

The authors state that the performance of the proposed cross layer optimizationis evaluated through simulations, MATLAB, LINDO API to solve related LP prob-lems, generic DTnet topology. The capacity of a wavelength assumed equals to10Gbps and finally to evaluate the feasibility of lightpath physical layer evaluationmodule was developed.

The authors state that assuming 16 wavelengths and random 100 traffic matrixthe proposed IA-RWA algorithm manages to serve all the traffic matric upto loadequals 0.8 with zero physical and networking blocking. For load equals 2.05 andfor wavelengths=35 the pure RWA algorithm has blocking equal to 10 percentagewhile the IA-RWA has blocking equal to 0 percentage and running time was about20min.

The authors claim that the cross layer optimization minimizes the number ofwavelength used in the network and also select lightpaths with acceptable qualityof transmission. The proposed algorithm has better performance improvementwhen compared to typical RWA algorithm.

This paper is cited by Aparicio-Pardo et al. [2011].

2.1.3 Noval Integer Linear Programming and Heuristic approach. Zhang et al.[2009] propose this approach to provide efficacious and productive way of transfer-ring data with less bit errors during transmission. The authors state that most ofresearch done in designing and routing optical signals assumes that the mediumin which the data is transferred passes without any error in bits. But in fact thesituation of non-erroneous transfer of data may not really occur since there arechances that some physical impairment may be introduced by the optical fibersor other components utilized to route optical signals. The authors state that suchimpairments should be taken into consideration while designing and routing opticalsignal and there should be appropriate techniques to transfer data with the help ofoptical signals without any bit errors during transmission and look into providingan efficient and effective routing of optical signal.

The authors do not refer to any previous work. However, they do state that eventhough physical impairments are considered during transmission of optical signals,there is no assurance of a particular request being accepted in order to transmitdata because the signal could possibly generate high bit error rate and requestscould be rejected. Hence many optical signal requests can not be accepted becauseof chances that it could not keep up the quality of the signal due to impairments,



which makes the whole network inefficient to accept many source and destinationrequests. Therefore there should be techniques to convert the inefficient signal toefficient signal even though some cost is involved so that we can increase the overallnetwork utilization up to its maximum capability.

The authors state that during transmission of an optical signal, impairments arecaused by the optical fibers or components used in the optical network which in turnis responsible for bit errors. To overcome the problem of bit errors during transmis-sion of optical signals the authors introduce a novel Integer Linear Programming(ILP) formulation for small networks and a heuristic approach for a large networkwhich places minimum number of 3R regenerators (where each R= reshaping, R=retiming, R= reamplification) in an efficient and effective way with optical electricaloptical conversion capability to accommodate all requests.

The authors state that in order to compare and evaluate the performance of theproposed methods the pacific bell network topology was used to test the networkwhich has 15 nodes and 21 links with 32 wavelengths per fiber capability apartfrom that CPLEX was also used to solve the ILP formulation to generate an optimalsolution. The simulation was performed on a machine with a Intel Pentium 1.70GHzCPU and 512 MB memory.

The authors state that for traffic matrix size upto 70 starting from 50 size, nearly4 regeneration nodes were required for ILP formulation method and for heuristicmethod it was varying from five to seven and beyond 70 traffic matrix size, ILPcould not generate solution. In case of the heuristic approach it generated a feasiblesolution as the traffic matrix size increases. The running time performance for ILPto generate solution was 1000 millisecond whereas for a heuristic to generate asolution it was taking a few milliseconds.

The authors claim that the results obtained from the experiments show thattheir heuristic algorithm outperforms the ILP optimal solution for various trafficmatrices and showed better result and performance than the ILP optimal solution.

2.1.4 Integer Linear Programming and Polynomial Time Heuristic approach.Sengezer et al. [2010] proposed this method to provide adequate use of bandwidthallocated while preventing any bit errors during transmission of lightpath request.The authors state that there is problem in establishing static versions of lightpathrequests which consider physical impairments in the optical layer which is the back-bone of the lower layer in the multilayer architecture which consists of a Servicelayer, Electrical layer and Optical layer. The authors state that there is a needto formulate efficient algorithms in order to satisfy maximum lightpaths (requests)with minimum blocking and bit error rates of the established lightpaths and makeefficient use of the bandwidth allocated for transmission of data.

The authors refer to previous work by Christodoulopoulos et al. [2009] andidentify a few shortcomings of that work, and of work by other authors also in thisfield of research working on the same problem. The authors state that the existingmethod was not investigated in detail and there was only a basic comparison withother methods which do not consider impairments and no further study was done.The authors state that there are many other factors which influence a lightpath tosatisfy a demand and must be taken into consideration and investigated further.

The authors propose an Integer Linear Programming (ILP) formulation for static



lightpath establishment which considers physical layer impairments. The idea be-hind the ILP formulation is to route as many lightpath demands by making useof the resources provided. In formulating the ILP the authors considered the ag-gregate effect of the physical impairments through which low Bit Error Rate isguaranteed which is below the acceptable threshold. A set of paths for each source-destination are generated and each request is routed through paths chosen and atevery level a link capacity constraint is imposed. A physical impairment model isused to provide with upper limit of impairments a lightpath can handle when it isrouted from the source, and the level at the receiver is monitored using a linearexpression which represent the noise level due to ASE, Cross-talk and by the signalitself. The authors state that there is a drawback in the ILP formulation as the sizeof the network increases the solution time also increases exponentially, which makesit infeasible for a large size network. To overcome this drawback a polynomial timeheuristic approach called ROLE (ReOrdered Lightpath Establishment Algorithm)was formulated taking into consideration large size networks the objective of theheuristic is to route as many requests as possible taking consideration of the BitError Rate constraint and not violating it. The authors formulated the ROLEheuristic in three phases first is Routing and Wavelength Assignment, reroutingand reordering. In Routing and Wavelength Assignment shortest path is selectedalong with an available wavelength and best combination is looked for a particu-lar lightpath demand. After RWA rerouting is performed in order to improve theresults of the RWA phase, which has a few lightpaths blocked due to no availablewavelength or due to BER threshold. The algorithm first tries to route the BERblocked demands and then wavelength blocked demands so that minimum nodecross talk is achieved. In case of reordering the authors state that all the unsatis-fied lightpath requests are made to undergo RWA and rerouting until there are nounsatisfied lightpath demands.

The authors state that in order to evaluate the performance of the ROLE al-gorithm proposed, and to make comparison with other approaches as well as tocompare with itself based on sorting techniques, a randomly generated physicaltopology each with 20 nodes having 25-40 bidirectional links ( topology1, topol-ogy2, topology3, topology4) using three different wavelengths 4,8,16 and whichhas the required lightpath demands generated randomly sorted in either shortestlightpath demand, longest lightpath demand was utilized. Experiments were alsocarried out on realistic Deutsche Telekom network topology in order to make avalid comparison with other proposed procedures. The ILP algorithm proposed bythe author was solved using ILOG CPLEX optimizer which uses branch and cutmethod was compared with ROLE for a very small network and performance wasevaluated.

According to the authors experiments on ROLE algorithm with taking into con-sideration both shortest demand first and longer demand first on various topologiesshows percentage of routed lightpaths is from 65 for topology1 to 87 for topology4.

The authors claim that the proposed ROLE algorithm performed 14 percentagebetter than the best heuristic algorithm present so far (POLIO-RWA) for the sameproblem, and is also guaranteed to provide optimum solutions for different sets ofdemands for different network sizes.



This paper is cited by Sengezer et al.[2012].

2.1.5 Integer Linear Programming and 3-Step,Lightpath Segmentation Heuris-tic approach. Aparicio-Pardo et al.[2011] propose this method to provide minimalinterference of lightpaths, wavelength conversion in order to avoid huge capital ex-penditure. The authors state that physical layer impairments degrade the qualityof optical signals during their transmission which cause the transmitted data notto be received correctly because of high bit error rate. There should be techniqueswhich consider these physical impairments and avoid or minimize the lightpathblocking, interferences between lightpaths, and wavelength conversion and improv-ing the quality of signal.

The authors state that the existing approach which considers nonlinear variationof impairment (caused due to neighboring lightpaths in the networks) minimizingthe bit error rate by placing regenerators and routing the signal according to thedemand still has some demands blocked due to deteriorating quality of transmissionand performance. In addition the authors do not consider the linear impairmentscaused due to the same lightpath. Therefore the authors state that the problemneeds to be solved by considering essential parameters which account for successfultransmission of optical signal and also satisfying maximum demands (requests).

The authors propose an Integer Linear Programming model which considers phys-ical layer linear impairments and which the authors claim to be a novel approachwhich was previously not done by any other researchers for a small or mediumsized network. Apart from the ILP model the authors also propose a differentheuristic approaches for a wide or large network which considers linear and non-linear physical impairments individually. The authors state that the technique oridea behind Integer Linear Programming was for each lightpath demand a set ofvalid semilightpaths which are optical signal traversing a sequence of fibers withoutundergoing any signal regeneration is calculated these semilightpaths forms a reach-ability graph which describes the ability of the semilightpaths to reach from onenode to another, to improve the strength of the signal minimum number of regener-ators planned are placed in order to improve and minimize the number of lightpathswhich are blocked. This technique works for a small or medium size network andnot for wide or large network, a new technique was proposed for such networkswhich is a heuristic approach and named it as Lightpath Segmentation, 3- StepHeuristic. The authors state that in Light path segmentation each candidate lightpath is divided or segmented into number of semilightpaths if the quality of signalmeasured by linear QoT estimator is less than the threshold, the degradation iscaused by the physical layer linear impairments. The signals are made transparentby placing regenerators on the nodes which requires signal regenerations reducingthe number of blocked light path requests, the algorithm minimizes such regener-ator placement on nodes requiring signal regeneration and maximizes the numberof lightpath requests from source to required destination. Here in order to makethe formulation simple not all the valid semilightpaths are considered only few areconsidered and restriction is imposed. The authors state that a second heuristicapproach called 3-step heuristic was also proposed which considers linear physicalimpairments in a large sized network. In this approach the authors initially allocateall the lightpaths without taking into consideration physical impairment, then check



whether wavelength continuity is not satisfying and signal degradation is occurringif yes then place regenerators accordingly for wavelength conversion and improvingthe quality of the signal to avoid bit errors and also to minimize network blockingto allocates maximum number of lightpath requests. The authors propose anotherheuristic called Iterative Regenerator Placement (IRP) which considers nonlinearphysical impairments and which uses nonlinear QoT estimator to evaluate the qual-ity of lightpath or semilightpaths used. Initially all the lightpaths are routed andwhere the wavelength continuity is not satisfying wavelength converters are placedthen an iterative regenerator placement is followed, through which semilightpathsapproaching a particular nodes having worst Q factor has a regenerator placed onthe node which has this as an outgoing semilightpath and along with it wavelengthis also assigned this is an iterative process and continues until all the resultingsemilightpaths have a valid Q factor along the signal.

The authors take into consideration three reference network topologies in orderto study the technique or approach proposed these reference topologies are INTER-NET, NSFNET and NOBEL-EU, for a medium size network which is INTERNET,NSFNET 8, 16 wavelengths were tested per fiber and 80 wavelengths were testedin case of NOBEL-EU. The authors utilize Q-Personick factor in order to repre-sent linear and non-linear QoT estimation, which considers 17 dB as maximumthreshold of the acceptable Q-factor, which is also used as reference to validate thequality of a semilightpath which has a maximum link length as 2688 Km. In orderto test authors consider a network topology, three traffic loads matrix which arelow, medium, high which has traffic load factor as 0.4, 0.7, 1, with certain wave-lengths per fiber the maximum number of lightpath demands are calculated andlater the effects of network link lengths on the regenerator equipment cost plannedare tested.

The authors state that the heuristics LS and Three-Step algorithms provide simi-lar optimal solution and in specific LS algorithm provided better solution comparedto rest algorithms as it requires less number of regenerators to be placed along withalso having minimum lightpath blocking capability when considering linear impair-ments. By observing the results of the three different algorithms which considerlinear impairments only it can be said that small sized networks do not requireregeneration equipment at all and it is topological dependent. Drawing attentiontowards the execution time for the proposed algorithms which takes into consider-ation linear impairments and comparing among it the authors state that it can beobserved that three-step algorithm has execution time below 1 second and highestexecution time was observed in the ILP formulation method but LS showed vari-ation in the execution time, it was less than one second in Internet and NSFNETnetwork topologies most of the cases of inputs provided and it increased with in-crease in network size. The authors also conducted what they claim to be exhaus-tive testing taking into consideration non-linear impairments using iterative routingplacement algorithm and observed that it outperforms ILP which considers physicalhops (PH-ILP) and also has a better execution time.

The authors claim that the non-linear IA-RWA-RP algorithm is the best, in bothlightpath blocking and regenerator cost, compared to PH-ILP and whereas in linearIA-RWA-RP the LS and three-step heuristics provide optimal or close to optimal



solutions in all tests conducted. In addition, the three-step algorithm guaranteethat no signal regeneration blocking is produced

This paper is cited by Sole et al.[2012].

2.1.6 Integer Linear Programming and Heuristic approach. Katrinis et al.[2011]propose this method to avoid give and take between cost of signal regeneration andeffectiveness of the signal and improve the ability of optical signal reaching longdistance. The authors state that the quality of a signal degrades as it propagatesthrough the optical medium and there should be techniques which take into accountimpairments caused due to physical layer during transmission, ways to strengthenthe signal quality, and methods to determine how many signals should be deployedin a network in order to avoid signal degradation. The authors also state thattechniques developed should be of low cost with an ability of signals to reach longdistances without compromising signal quality and avoid any tradeoff between ef-fectiveness of signal and cost of signal regeneration.

The authors state certain drawback from the previous approaches proposed inthe same area of problem and state that with the existing techniques there wasclear tradeoff between the number of regenerators to be placed and the capacityof the WDM network. There was more concentration on routing aspects ratherthan design of the network which is the most important before routing any lightpath demand. The authors also state that the existing methods assumed that thedistance covered by lightpath is a good approximation of signal quality degradationwhich is not necessarily true and further it holds good only for cases where thereare limited number of hops and for networks exhibiting high homogeneity of opticalequipment.

The authors state that a new Integer Linear Programme was formulated, whichaccording to authors was not previously stated by others in the literature and theyconsider it to be a novel approach. The authors state that the main aim of theirILP formulation is to specify the capacity (which is the number of fibers), numberof links (Edges) through which a candidate path travels, number of wavelength perfiber it supports and minimize the total cost occurring of the dimensioned (theoptimization of the capacity put on deployed network) network.

The authors state that inorder to test the ILP formulation proposed, the ILOGCPLEX v.11.1.1 mathematical programming engine was used for solving the IntegerLinear Program instances. Impairment models were used to evaluate the Bit ErrorRate throughout the experiment. The hardware utilized during the experimentwas Intel Xenon CPU 2.80 GHz Linux workstation, equipped with 2GB RAM. Thenetwork topology utilized was NSFNET comprising of 16 nodes and 48 physicallinks, in regards to input traffic requests a randomly generated traffic matrix with70 nonzero source/destination pairs, with 344 wavelengths was used, apart from itnoise figures of amplifiers which are modeled using a normal distributed randomvariable with certain standard deviation and step level was used for the purpose ofevaluation of optical equipment configuration.

The authors state that for optical reach values from 500 to 1250km and low het-erogeneity (variation in optical equipment) percentage of impaired lightpaths wasfrom 0 to 2.5 percentage for network dimensioning in combination with length-basedregeneration. As the degree of heterogeneity increased the percentage of impaired



lightpaths increased from 2.5 percentage to 14.6 percentage. For optical reach 1500km and 1750 km the percentage of impaired paths went upto 21 percentage. Theauthors also state that 90 percentage of saving in regenerators with IA Regener-ator placement approach was seen for optical reach which is 500 and for variousheterogeneity levels and then for optical reach 1750 the impairment aware regenera-tor placement has 60 percentage saving in regenerators used. When considering thecase of uncertainty of heterogeneous levels impairment aware regenerator placementhas on average of 138-209 regenerators used on average, specific to heterogeneitylevel 5 the saving was 35 percentage on average and yielding 10 percentage impairedpaths on average.

The authors claim that the results obtained from the experiments show theirapproach has better results than the previous approach which considers placingregenerators based on the length rather than based on the quality of signal andachieved less regeneration cost along with providing good signal quality.

2.1.7 Mixed Integer Linear Programming and Tabu, Greedy Search Heuristic ap-proach. Sengezer et al. [2012] propose a method which allows information passingfrom one layers to another in a network. The authors state that there is a need forproper topology design through which we can obtain less or no wavelength conver-sion at all during transmission of an optical signal considering physical impairmentsalong its path. There is a need for design or a approach in which sharing of infor-mation between various layers in the wavelength division multiplexing network istaken care of in order to attain blocking-free performance and best utilization ofresources available.

The authors refer to work done by Varvarigos et.al [2009], Sengezer et.al [2010]and points out certain shortcomings from the previous approaches referred as wellas other techniques which are used to solve the problem and state that in theprevious approaches the solutions obtained was not traceable for moderate or largesize networks as the whole problem was divided into sub problems and there shouldalso be care taken during integration of solutions obtained at every phase and ifthere is a mistake then we would not obtain optimal solution at all. The authors alsostate that all the subproblems consider only one layer and omit other layers presentand ignore sharing of information also during its transmission in an optical networkand physical layer impairments are also ignored. A heuristic approach formulatedprovided solution but it was observed that excess utilization of wavelengths thanused was possible and further does not provide the optimal solution.

The authors state that their approach is the first of its kind in the area to solvethe problem. They introduce, what they claim to be, a novel multilayer virtualtopology design approach in which both wavelength assignment and physical layerimpairments are considered, and which involves interaction of control planes ofmulti-protocol label switching and wavelength division multiplexing layers duringtopology design and share information using topology information sharing strat-egy in which a wavelength routing node architecture was followed in physical layermodel. The authors also propose two topology designs: a) Tabu search virtualtopology design, and b) Greedy search virtual topology design in order to solvemultilayer virtual topology design. Apart from this a mixed integer linear program-ming approach was made to provide a bench mark to evaluate the performance of



other approaches to the multi-layer VTD problem.The authors state that in order to evaluate the performance of the VTD algo-

rithms and information strategies with different traffic rates and patterns changingwith time. A wide area network is utilized with NSFNET topology with 21 linksand 14 nodes spread over four different time zones. Apart from this inorder toevaluate the topology design algorithms CPLEX 12.1 optimization problem solverwas used.

The authors state that on conducting experiment with 15 percentage traffic mag-nitude with 21 links and 14 nodes MILP shows 31.9, GS-VTD shows 27.82, TS-VTDwith out layer interaction shows 36.7 and TS-VTD with layer interaction shows 0.03percentages of bandwidth blocking and for a whole day time period blocking ra-tio for a single-hour traffic demand was 3.19 percentage and for multi-hour trafficdemand it was 2.40 percentage both having 100 percentage traffic intensity.

The authors claim that the proposed method solves the multi-layer logical topol-ogy design and LSP routing problems arising in MPLS layers, and routing andwavelength assignment problems in the wdm layers in an integrated approach thatdoes not necessitate a single common control plane for both layers. The authorsalso state that MILP formulation approach does not generate good solutions andwhen taking into consideration heuristic approach Tabu search based algorithm canreduce the blocking ratio of the GS-VTD algorithm by 35 to 100 percentage.



2.1.8 Summary.

Year Author Title of Paper Major Contribution2009 Manousakis et

al.OfflineImpairment-AwareRouting and Wave-length AssignmentAlgorithms inTranslucent WDMOptical Networks

A novel integer linear pro-gramming and heuristicapproach for routing andwavelength assignment alongwith solving regeneratorplacement.

2009 Pavon-Marinoet al.

Offline ImpairmentAware RWA Algo-rithms for Cross-Layer Planning ofOptical Networks

An integer linear program-ming and heuristic approachwhich perform cross layer op-timization serving maximumlightpath requests

2009 Zhang et al. REPARE: Regener-ator Placement andRouting Establish-ment in Translu-cent Networks

A novel Integer Linear Pro-gramming formulation andheuristic approach for placingregenerators.

2010 Sengezer et al. Static LightpathEstablishment inMultilayer TrafficEngineering UnderPhysical LayerImpairments

An integer linear program-ming and heuristic approachwhich consider aggregate ef-fects of physical impairmentsand guarantee low bit errorrate.

2011 Aparicio-Pardoet al.

OfflineImpairment-AwareRWA and Regen-erator Placementin TranslucentOptical Networks

A novel Integer Linear Pro-gramming and heuristic ap-proach which minimize light-path interference and block-ing.

2011 Katrnis et al. On the Dimension-ing of WDM Opti-cal Networks WithImpairment-AwareRegeneration

A novel integer linear pro-gramming and heuristic ap-proach which helps in pre-venting trade-off between ef-fectiveness of signal and costof signal regeneration andwhich minimizes the totalcapital expenditure cost.

2012 Sengezer et al. Multi-layer VirtualTopology Design inOptical NetworksUnder PhysicalLayer Impairmentsand Multi-hourTraffic Demand

An integer linear program-ming and heuristic approachto design a novel multilayervirtual topology.



2.2 Integer Linear Programming or Heuristic Based approaches

The three Research papers that are considered in this section use Integer Lin-ear Programming (ILP) or Heuristic based approaches to solve the problem. Allof this research adopts an ILP or a Heuristic approach to route lightpath re-quests under physical layer impairments. An initial approach was proposed byChristodoulopoulous et al. [2009] by extending the previous approach (RWA whichdo not consider impairments) and is followed by other approaches by various au-thors which are either new, extension or alteration to an existing technique andwhich focus on a small network and tries to provide a close to optimal solution torealistic problem.

2.2.1 Integer Linear Programming approach. Christodoulopoulos et al.[2009]state that certain physical effects which are sources of impairments are not takeninto consideration during routing and wavelength assignment of lightpath to trans-fer data from source to destination. These impairments affects selection of light-paths on itself as well as other lightpaths adjacent to it during routing and foundthat it was difficult for them to formulate in an offline algorithms. They state thatthere should be technique to incorporate such impairments an algorithm duringrouting and wavelength assignment so that experiment can be done which are ex-actly behaving as in real time scenario and come up with a solution which helpsin minimizing the number of wavelengths used and also select lightpaths that havebetter quality of transmission.

The authors do not refer to any previous work. However, they identify drawbackfrom the research previously done and state that physical layer impairments werenot considered during routing of lightpaths.

The authors introduce the IA-RWA(Worst CaseIA-RWA) algorithm by extend-ing the pure routing wavelength assignment algorithm taking into consideration thephysical impairments which are generated by presence of other lightpaths or thosethat affect the same lightpath that generated them (assuming worst case scenario:where all lightpaths are active), using analytical model the effects of physical layerimpairments for each candidate lightpath are calculated and only those which haveacceptable quality of transmission are taken into consideration to transfer data andrest are ignored. The authors propose another algorithm which is a cross layerimpairment aware routing wavelength assignment (sigma bound IA-RWA) the au-thors initially proceed by constraining the total number of adjacent, second adjacentchannels and intra-XT interfering sources on a particular lightpath using mathe-matical methods which accounts for impairments that affect the same lightpaththat generated them, the authors then consider noise variance which accounts theeffects of the impairments caused by presence of other lightpaths. In sigma boundIA-RWA algorithm for a given lightpath after obtaining the number of adjacentchannel and interfering sources for a particular lightpath it is multiplied with thenoise variance and only those lightpaths that satisfy the corresponding noise vari-ance constraints are the one which has acceptable quality of transmission and areused to transfer data.

The authors carried out different simulation experiments and implemented pureRWA and the proposed IA-RWA in Matlab and to evaluate the feasibility of light-path Q-Tool developed with in the DICONET project was used. The topology used



was the Deutsche Telekom network with a capacity of each wavelength channel is10 Gb/s.

The authors present the results obtained after conducting the experiments arepresented in the form of histograms which graphs distribution of adjacent channelsand number of intra XT interfering sources, and observed that there is a left shift inthe probability distribution with sigma bound IA-RWA when compared with pureRWA which does not consider impairments. This indicates that the impairmentsgenerating sources are less and the quality of signal is more for a sigma boundIA-RWA compared to RWA. Moreover Sigma Bound-IA-RWA surpasses WC-IA-RWA by achieving zero blocking with few wavelengths and also with less numberof rejected request calls for transmission of data.

The authors state that SB-IA-RWA gives better performance than the pure RWAalgorithm which does not take into consideration the physical impairments or worstcase IA-RWA which assumes that all lightpaths are active.

This paper is cited by Sengezer et al. [2010].

2.2.2 Relaxed Integer Linear Programming approach. Varvarigos et al. [2010]The authors state that signal quality degrades during transmission from source todestination due to physical layer impairments and also effects the routing choicesmade for a particular lightpath as well as other lightpaths which are transmittedalong the same fiber. The authors state that there should be certain optimizationtechniques which help in providing connections by utilizing a minimum number ofwavelengths but also helps in providing quality signal which is acceptable.

The authors do not to any previous work but identify shortcomings of previ-ous approach by other authors working on the same problem. The authors statethat decomposition techniques were followed in previous approaches which mightnot provide an optimal solution for the joint Routing and Wavelength Assignmentproblem, Apart from this, when taking into consideration interference, existingapproaches do not consider the interference due to already established lightpathson the candidate lightpath. Instead they assume a worst case scenario where allthe lightpaths are active. The authors state that a new optimal technique shouldbe formulated in order overcome existing problems and shortcoming of previousapproaches.

The authors state that an initial approach was developed to solve the Routingand Wavelength Assignment problem without considering impairments based onLinear Programming relaxation formulation in order to generate integer solutions,then the algorithm is extended to two impairment aware routing, and a wave-length assignment algorithm was proposed that accounts for interference withinlightpaths. The authors state that one approach takes physical impairments di-rectly into account with the help of utilizing noise related parameters, whereas theother approach takes into effect the physical impairments indirectly and formulatea cross layer approach and helps in satisfying maximum number of requests (sourceto destination) with acceptable signal quality until it reaches the destination

The authors state that in order to test the proposed algorithms numerous simula-tion experiments were carried out in Matlab with random static instances. LINDOwas used in order to solve the LP and ILP problems and the network topology usedin their simulations was the generic deutsche Telekom network and the capacity



of the wavelength is considered to be 10 Gb/s. In order to evaluate the feasibilityof the lightpaths a QoT tool was used with acceptable Q-factor equal to 15.5 dB.For evaluating the IA- RWA algorithm a link model with NRZ-OOK modulationformat, 10-gb/s transmission rates and 50 GHz channel spacing with span lengthof 100km each having SSMF fibers with dispersion parameter of 17 ps/nm/km andattenuation of 25 db/km, was used.

The authors state that with traffic matrix load between 0.5 and 3 as input. TheSB-IA-RWA scales well with the load. Considering realistic traffic matrix of DTnetin the simulation experiments the IA-RWA blocking ratio decreases significantly.The SB-IA-RWA can obtain a zero blocking solution, with running time about15 min utilizing around 19700 variables and 22515 constraints for 35 wavelengths,about 3.2 percentage and running time around 3 hours utilizing 21338 variablesand 61278 constraints for P-IA-RWA and 10 percentage with pure RWA and withrunning time around 30 seconds.

The authors claim that the approach of solving RWA problem based on LP-relaxation generates an optimal solution despite absence of integrality constraintsfor large instances. The authors further claim that the proposed two IA RWA al-gorithms which are P-IA-RWA and SB-IA-RWA show better performance than thepure RWA algorithm and moreover SB-IA-RWA shows better performance minimiz-ing wavelength utilization providing zero-blocking solution with very less executiontimes.

2.2.3 Heuristic, Meta-Heuristic or Combinatorial approach. Sole et al. [2012]state that in a transparent network which transmits data in optical mode from endnode to end node (thus reducing the expenses of converting the signal from opticalto electrical and then to optical mode) there are chances of physical impairments ac-cumulating on the signal during transmission which reduces the optical reach of thesignal in the network and is also the source for other failures. The authors state thatproper planning and operation phases should be taken care of in advance in orderto avoid such problems, and there should also be certain techniques incorporatedwhich take care of both physical and network layer interaction and which enableproper allocation of the resources and requirements so that optimal solution is ob-tained with maximum allocation of demands. In addition to this, the authors statethat there should also be certain techniques which take into consideration integra-tion of measurements of optical layer performance into algorithms which considersimpairments on the lightpath signals during routing and wavelength assignment foroptical networks to be capable of being automated and network configurable.

The authors refer to previous work by Aparicio-Pardo et.al [2011], and Varvarigoset.al [2009]

The authors state that change in currently used fiber capacity from several 10-gigs of data to several 100-gigs or 400-gigs of data poses technical problems andexisting strategies does not apply to such wide capacity of fiber and these are yetto be investigated

The authors state that a cross layer strategy was developed which considersphysical layer information, to overcome the problems under investigation.

The authors state that all the strategies developed were investigated with a10gigs/sec system and infrastructure.



The authors state that there were no raw results obtained as no comparison ofthe different algorithm were carried out.

The authors claim that good Impairment Aware RWA algorithms were devel-oped for protected and unprotected lightpaths with optimal monitor placementtechniques.

2.2.4 Summary.

Year Author Title of Paper Major Contribution2009 Christodoulopoulous

et al.Considering phys-ical layer impair-ments in offlineRWA

Extended pure routing wave-length algorithm which con-siders impairments and pro-vided a algorithm which isclose to realistic scenario.

2010 Varvarigos et al. Offline Routingand WavelengthAssignment inTransparent WDMNetworks

An optimization techniquewhich utilize minimum wave-lengths and provide accept-able quality signal.

2012 Sole et al. Cross-Layer Ap-proaches for Plan-ning and OperatingImpairment-AwareOptical Networks

Technique which integratesoptical layer performancewhich makes optical networkautomated and networkconfigurable.

3. CITED BY

Author Cited byChristodoulopoulous et al. [2009] Sengezer et al. [2010]Manousakis et al. [2009] Aparicio-Pardo et al. [2011] and Sole et

al. [2012].Pavon-Marino et. al [2009] Aparicio-Pardo et al.[2011]Sengezer et al. [2010] Sengezer et al.[2012]Aparicio-pardo et al. [2011] Sole et al.[2012]

4. CONCLUDING COMMENTS

We note that Christodoulopoulos’s group has conducted significant research in thisarea and is also referred by many other research groups, which shows the qualityof their work.

Many of the approaches followed QoT threshold rather than taking distance i.e.optical reach where the signal tends to degrade its quality and errors in bits startsbuilding up.

Many authors first explain the simple network where in most of the cases, theregular RWA involving Integer Linear Programming approach solves the problemand shows better performance, and then later on, extends to solve large or mediumnetworks with the help of heuristic approach.



Most of the researchers used CPLEX in order to solve the integer linear program-ming and was given preference over other optimizers present.

All the researchers conducted experiments on standard real time networks presentand did not vaguely or randomly formulate one.

The amount of data transferred per fiber ranges up to around 10 gigs per fiberand none of the researchers have considered a fiber carrying 100gigs/400 gigs ormore. Imagine a collection of such fibers and each carrying 100gigs of data, fewsuch fibers crossing a node leads to a real cost effective solution as with low costlarge data can be transmitted. Taking into consideration the future scenario ofincrease in number of users this could help by carrying enormous amount of datathrough a fiber and make the communication quick and easy and have a low cost.

5. ACKNOWLEDGEMENT

I have the honour to acknowledge the knowledge, proficiency, expertise, judgmentand guidance I received from Dr. Richard Frost throughout this survey. I conveymy deep salutation to my supervisor Dr. Subir Bandyopadhyay for encouraging me,helping me throughly by providing me huge strength and support in completing thissurvey report and finally I would also like to thank Dr. Quazi Rahman, and SajibSinha for their time and continuous support through out my studies at Universityof Windsor.

6. ANNOTATIONS

6.1 Aparicio-Pardo et al. 2011

Citation:

Aparicio-Pardo, R., Klinkowski, M., Garcia-Manrubia, B. , Pavon-Marino,P., and Careglio, D. 2011. Offline Impairment-Aware RWA and RegeneratorPlacement in Translucent Optical Networks. Journal of Lightwave Technology 29,3,26–277.Problem: The authors state that physical layer impairments degrade the quality

of optical signals during their transmission which cause the transmitted data notto be received correctly because of high bit error rate. There should be techniqueswhich consider these physical impairments and avoid or minimize the lightpathblocking, interferences between lightpaths, and wavelength conversion and improv-ing the quality of signal.Previous Work: The authors refer to work done by K.Christodoulopoulos et al.

[2009].Shortcomings of Previous Work: The authors point out certain shortcomings

from the paper referred and also from previous approaches on the same problem.The authors state that the existing approach which considers nonlinear variationof impairment (caused due to neighboring lightpaths in the networks) minimizingthe bit error rate by placing regenerators and routing the signal according to thedemand still has some demands blocked due to deteriorating quality of transmissionand performance. In addition the authors do not consider the linear impairmentscaused due to the same lightpath. Therefore the authors state that the problemneed to be solved by considering essential parameters which accounts for successfultransmission of optical signal with maximum demands (requests) satisfying.



New Idea/Algorithm/Architecture: The authors propose an Integer Linear Pro-gramming model which considers physical layer linear impairments and which theauthors claim to be a novel approach which was previously not done by any otherresearchers for a small or medium sized network. Apart from the ILP model theauthors also propose a different heuristic approaches for a wide or large networkwhich considers linear and non-linear physical impairments individually. The au-thors state that the technique or idea behind Integer Linear Programming was foreach lightpath demand a set of valid semilightpaths which are optical signal travers-ing a sequence of fibers without undergoing any signal regeneration is calculatedthese semilightpaths forms a reachability graph which describes the ability of thesemilightpaths to reach from one node to another, to improve the strength of thesignal minimum number of regenerators planned are placed in order to improveand minimize the number of lightpaths which are blocked. This technique worksfor a small or medium size network and not for wide or large network, a new tech-nique was proposed for such networks which is a heuristic approach and namedit as Lightpath Segmentation, 3- Step Heuristic. The authors state that in Lightpath segmentation each candidate light path is divided or segmented into numberof semilightpaths if the quality of signal measured by linear QoT estimator is lessthan the threshold, the degradation is caused by the physical layer linear impair-ments. The signals are made transparent by placing regenerators on the nodeswhich requires signal regenerations reducing the number of blocked light path re-quests, the algorithm minimizes such regenerator placement on nodes requiringsignal regeneration and maximizes the number of lightpath requests from source torequired destination. Here in order to make the formulation simple not all the validsemilightpaths are considered only few are considered and restriction is imposed.The authors state that a second heuristic approach called 3-step heuristic was alsoproposed which considers linear physical impairments in a large sized network. Inthis approach the authors initially allocate all the lightpaths without taking intoconsideration physical impairment, then check whether wavelength continuity is notsatisfying and signal degradation is occurring if yes then place regenerators accord-ingly for wavelength conversion and improving the quality of the signal to avoidbit errors and also to minimize network blocking to allocates maximum number oflightpath requests. The authors propose another heuristic called Iterative Regen-erator Placement (IRP) which considers nonlinear physical impairments and whichuses nonlinear QoT estimator to evaluate the quality of lightpath or semilightpathsused. Initially all the lightpaths are routed and where the wavelength continu-ity is not satisfying wavelength converters are placed then an iterative regeneratorplacement is followed, through which semilightpaths approaching a particular nodeshaving worst Q factor has a regenerator placed on the node which has this as anoutgoing semilightpath and along with it wavelength is also assigned this is an it-erative process and continues until all the resulting semilightpaths have a valid Qfactor along the signal.

Experiments Conducted: The authors take into consideration three reference net-work topologies in order to study the technique or approach proposed these refer-ence topologies are INTERNET, NSFNET and NOBEL-EU, for a medium sizenetwork which is INTERNET, NSFNET 8, 16 wavelengths were tested per fiber



and 80 wavelengths were tested in case of NOBEL-EU. The authors utilize Q-Personick factor in order to represent linear and non-linear QoT estimation, whichconsiders 17 dB as maximum threshold of the acceptable Q-factor, which is alsoused as reference to validate the quality of a semilightpath which has a maximumlink length as 2688 Km. In order to test authors consider a network topology,three traffic loads matrix which are low, medium, high which has traffic load factoras 0.4, 0.7, 1, with certain wavelengths per fiber the maximum number of light-path demands are calculated and later the effects of network link lengths on theregenerator equipment cost planned are tested.

Results: The authors state that the heuristics LS and Three-Step algorithmsprovide similar optimal solution and in specific LS algorithm provided better solu-tion compared to rest algorithms as it requires less number of regenerators to beplaced along with also having minimum lightpath blocking capability when consid-ering linear impairments. By observing the results of the three different algorithmswhich consider linear impairments only it can be said that small sized networks donot require regeneration equipment at all and it is topological dependent. Drawingattention towards the execution time for the proposed algorithms which takes intoconsideration linear impairments and comparing among it the authors state thatit can be observed that three-step algorithm has execution time below 1 secondand highest execution time was observed in the ILP formulation method but LSshowed variation in the execution time, it was less than one second in Internet andNSFNET network topologies most of the cases of inputs provided and it increasedwith increase in network size. The authors also conducted what they claim to beexhaustive testing taking into consideration non-linear impairments using iterativerouting placement algorithm and observed that it outperforms ILP which considersphysical hops (PH-ILP) and also has a better execution time.Claims: The authors state that the non-linear IA-RWA-RP algorithm is best, in

both lightpath blocking and regenerator cost, compared to PH-ILP and whereasin linear IA-RWA-RP the LS and three-step heuristics provide optimal or closeto optimal solutions in all tests conducted. In addition, the three-step algorithmguarantee that no signal regeneration blocking is producedCitations by Others: Sole et al.[2012]..

6.2 Christodoulopoulos et al. 2009

Citation:

Christodoulopoulos, K., Manousakis, K., and Varvarigos, E., 2009. Con-sidering Physical Layer Impairments in Offline RWA. Network, IEEE 23, June,26–33.Problem: The authors state that certain physical effects which are sources of

impairments are not taken into consideration during routing and wavelength as-signment of lightpath to transfer data from source to destination. These impair-ments effects selection of lightpaths on itself as well as other lightpaths adjacentto it during routing and found that it was difficult for them to formulate in anoffline algorithms. They state that there should be technique to incorporate suchimpairments in an algorithm during routing and wavelength assignment so thatexperiment can be done which are exactly behaving as in real time scenario and



come up with a solution which helps in minimizing the number of wavelengths usedand also select lightpaths that have better quality of transmission.Previous Work: There was no previous work referred by authors but in general

state certain drawback from the research previously done in the same field.Shortcomings of Previous Work: The authors state that physical layer impair-

ments were not considered during routing of lightpaths.New Idea/Algorithm/Architecture: The authors introduce the IA-RWA(Worst

CaseIA-RWA) algorithm by extending the pure routing wavelength assignment al-gorithm taking into consideration the physical impairments which are generated bypresence of other lightpaths or those that affect the same lightpath that generatedthem (assuming worst case scenario: where all lightpaths are active), using ana-lytical model the effects of physical layer impairments for each candidate lightpathare calculated and only those which have acceptable quality of transmission aretaken into consideration to transfer data and rest are ignored. The authors pro-pose another algorithm which is a cross layer impairment aware routing wavelengthassignment (sigma bound IA-RWA) the authors initially proceed by constrainingthe total number of adjacent, second adjacent channels and intra-XT interferingsources on a particular lightpath using mathematical methods which accounts forimpairments that affect the same lightpath that generated them, the authors thenconsider noise variance which accounts the effects of the impairments caused bypresence of other lightpaths. In sigma bound IA-RWA algorithm for a given light-path after obtaining the number of adjacent channel and interfering sources for aparticular lightpath it is multiplied with the noise variance and only those light-paths that satisfy the corresponding noise variance constraints are the one whichhas acceptable quality of transmission and are used to transfer data.Experiments Conducted: The authors carried out different simulation experi-

ments and implemented pure RWA and the proposed IA-RWA in Matlab and toevaluate the feasibility of lightpath Q-Tool developed with in the DICONET projectwas used. The topology used was the Deutsche Telekom network with a capacityof each wavelength channel is 10 Gb/s.Results:The authors present the results obtained after conducting the experi-

ments are presented in the form of histograms which graphs distribution of adjacentchannels and number of intra XT interfering sources, and observed that there isa left shift in the probability distribution with sigma bound IA-RWA when com-pared with pure RWA which does not consider impairments. This indicates thatthe impairments generating sources are less and the quality of signal is more fora sigma bound IA-RWA compared to RWA. Moreover Sigma Bound-IA-RWA sur-passes WC-IA-RWA by achieving zero blocking with few wavelengths and also withless number of rejected request calls for transmission of data.Claims: The authors state that SB-IA-RWA gives better performance than the

pure RWA algorithm which does not take into consideration the physical impair-ments or worst case IA-RWA which assumes that all lightpaths are active.Citations by Others: This paper is cited by Sengezer et al. [2010].

6.3 Katrinis et al. 2011

Citation:



Katrinis, M. and Tzanakaki, A. 2011. On the Dimensioning of WDM OpticalNetworks with Impairment-Aware Regeneration IEEE/ACM Transactions 19, 3,735–746.Problem: The authors state that the quality of a signal degrades as it propagatesthrough the optical medium and there should be techniques which take into accountimpairments caused due to physical layer during transmission, ways to strengthenthe signal quality, and methods to determine how many signals should be deployedin a network in order to avoid signal degradation. The authors also state thattechniques developed should be of low cost with an ability of signals to reach longdistances without compromising signal quality and avoid any tradeoff between ef-fectiveness of signal and cost of signal regeneration.

Previous Work: The authors do not refer to any paper which is present in list ofpapers selected which are exactly on topic.

Shortcomings of Previous Work: The authors state certain drawback from theprevious approaches proposed in the same area of problem and state that with theexisting techniques there was clear tradeoff between the number of regenerators tobe placed and the capacity of the WDM network. There was more concentrationon routing aspects rather than design of the network which is the most importantbefore routing any light path demand. The authors also state that the existingmethods assumed that the distance covered by lightpath is a good approximationof signal quality degradation which is not necessarily true and further it holds goodonly for cases where there are limited number of hops and for networks exhibitinghigh homogeneity of optical equipment.

New Idea/Algorithm/Architecture:The authors state that a new Integer LinearProgramme was formulated, which according to authors was not previously statedby others in the literature and they consider it to be a novel approach. The authorsstate that the main aim of their ILP formulation is to specify the capacity (whichis the number of fibers), number of links (Edges) through which a candidate pathtravels, number of wavelength per fiber it supports and minimize the total costoccurring of the dimensioned (the optimization of the capacity put on deployednetwork) network.

Experiments Conducted:The authors state that inorder to test the ILP formula-tion proposed, the ILOG CPLEX v.11.1.1 mathematical programming engine wasused for solving the Integer Linear Program instances. Impairment models wereused to evaluate the Bit Error Rate throughout the experiment. The hardwareutilized during the experiment was Intel Xenon CPU 2.80 GHz Linux workstation,equipped with 2GB RAM. The network topology utilized was NSFNET comprisingof 16 nodes and 48 physical links, in regards to input traffic requests a randomlygenerated traffic matrix with 70 nonzero source/destination pairs, with 344 wave-lengths was used, apart from it noise figures of amplifiers which are modeled using anormal distributed random variable with certain standard deviation and step levelwas used for the purpose of evaluation of optical equipment configuration.

Results: The authors state that for optical reach values from 500 to 1250km andlow heterogeneity (variation in optical equipment) percentage of impaired light-paths was from 0 to 2.5 percentage for network dimensioning in combination withlength-based regeneration. As the degree of heterogeneity increased the percent-



age of impaired lightpaths increased from 2.5 percentage to 14.6 percentage. Foroptical reach 1500 km and 1750 km the percentage of impaired paths went upto21 percentage. The authors also state that 90 percentage of saving in regeneratorswith IA Regenerator placement approach was seen for optical reach which is 500and for various heterogeneity levels and then for optical reach 1750 the impairmentaware regenerator placement has 60 percentage saving in regenerators used. Whenconsidering the case of uncertainty of heterogeneous levels impairment aware regen-erator placement has on average of 138-209 regenerators used on average, specificto heterogeneity level 5 the saving was 35 percentage on average and yielding 10percentage impaired paths on average.Claims: The authors claim that the results obtained from the experiments show

their approach has better results than the previous approach which considers plac-ing regenerators based on the length rather than based on the quality of signal andachieved less regeneration cost along with providing good signal quality.

Citations by Others: There was no citation done by the authors from the list ofmost important papers referred on topic.

6.4 Manousakis et al. 2009

Citation:

Manousakis, K., Christodoulopoulos, K., Kamitsas, E., Tomkos, I., andVarvarigos, E. 2009. Offline Impairment-Aware Routing and Wavelength As-signment Algorithms in Translucent WDM Optical Networks. Journal of LightwaveTechnology 27, 12, 1866–1877.

Problem: The authors state that during transmission of an optical signal theexistence of physical impairments in optical fibers block the connection requestsfrom reaching the destination. There should be efficient algorithms to re-routeblocked connections by selective placement as well as selective use of minimumnumber of regenerators which re-shapes, re-amplify and re-time the signal.Previous Work: The authors refer to work of Manousakis et.al [2009].Shortcomings of Previous Work: The authors state no exact short comings in

the approach referred but points out the shortcomings in general taking the overall scenario and state that in the previous approaches proposed used regeneratorsat every node which do not provide a cost effective solution and is a huge drawbackand few other approaches do not consider the impairment effect at all.New Idea/Algorithm/Architecture:The authors state that a new impairment aware

algorithm was designed and implemented for routing and wavelength assignmentalong with solving regenerator placement and assignment problem in translucentnetworks which places regenerators at selective sites along with the number ofregenerators to be placed. The authors state that the placement and assigning ofregenerators is considered as a virtual topology design problem which utilize variousILP and greedy heuristic algorithms to solve it and apart from that the approachalso determines sequence of regenerators to be utilized by the connections havingimpairments and transform the non-transparent connection to transparent signalby terminating and regenerating the signal at the specified intermediate node andminimize the number of blocked connections.Experiments Conducted: The authors state that in order to evaluate the per-



formance of the proposed impairment aware RWA algorithm numerous simulationexperiments were performed with the IA-RWA algorithm. Virtual topology algo-rithms were implemented in Matlab and LINDO API was used to solve ILP relatedproblems in order to evaluate the feasibility of lightpath. The Q-factor estimatorwas used in the experiment. The link model has a NRZ-OOK modulation format,10 Gbps transmission rates, 50 GHz channel spacing with span length equals to80km with Geant-2 network topology with 34 nodes and 54 bidirectional links.

Results:The authors state that when the blocking ratio is graphed against thenumber of available wavelengths and demands the PH-based algorithms shows ablocking probability ranging from 0.05 to 0.00 and required wavelengths rangingfrom 90 to 110 to reach zero blocking with the number of demands varying from700 to 400. Moreover the PH-ILP sum of regenerator algorithm required 115wave-lengths and 461 regenerators while the VP-ILP sum of regenerator required 135and 430 regenerators to reach zero blocking. In addition, the execution time ofthe IA-RWA algorithm for translucent network requires about 3 hours in whichapplication of IA-RWA algorithm used by transparent network in translucent net-work consumes most of the time and the remaining phases requires only few 10s ofseconds.Claims: The authors state that with the use of the Q factor, low complexity

of the virtual topology design problem make the approach efficient for small andlarge scale heterogeneous mesh networks. Among various algorithms proposed forthe virtual topology design problem the ILP formulation helps in reducing themaximum number of regenerators to be placed and among the heuristics proposedthe physical Hop heuristic exhibit better performance and stable function in allcases.Citations by Others: This paper is cited by Aparicio-Pardo e al. [2011] and sole

et al. [2012].

6.5 Pavon-Marino et al. 2009

Citation:

Pavon-Marino, P., Azodolmolky, S., Aparicio-Pardo, R., Garcia-Manrubia,B., Pointurier, Y.,Angelou, M., Sole-Pareta, J.,Garcia-Haro, J., andTomkos, I. 2009. Offline Impairment Aware RWA Algorithms for Cross-LayerPlanning of Optical Networks. Journal of Lightwave Technology 27,12,1763—1775.

Problem: The authors state that the quality of signal degrades due to physicallayer impairment and if there are no regenerators there are chances of bit errorsto be more which in turn weakens the signal strength. The authors state thatthere should be appropriate techniques which consider impairments during signaltransmission and help in serving maximum number of connections

Previous Work: The authors state that there is no paper referred by the authors.Shortcomings of Previous Work: No shortcomings are mentioned by the authors.New Idea/Algorithm/Architecture: The authors state that an IA-RWA offline

algorithmic approach is followed which performs a cross layer optimization betweenthe physical and network layers present and take into account the directly thedominant impairment effect and upper bound on the interference noise variance itcan tolerate is calculated and it is used as constraint the interfering noise caused



by other lightpaths and avoid heavy ILP formulations.Experiments Conducted: The authors state that the performance of the proposed

crosses layer optimization is evaluated through simulations, MATLAB, LINDO APIto solve related LP problems, generic DTnet topology. The capacity of a wavelengthassumed equals to 10Gbps and finally to evaluate the feasibility of lightpath physicallayer evaluation module was developed.Results:The authors state that assuming 16 wavelengths and random 100 traffic

matrix the proposed IA-RWA algorithm manages to serve all the traffic matricupto load equals 0.8 with zero physical and networking blocking. For load equals2.05 and for wavelengths=35 the pure RWA algorithm has blocking equal to 10percentage while the IA-RWA has blocking equal to 0 percentage and running timewas about 20min.Claims:The authors claim that the cross layer optimization minimizes the number

of wavelength used in the network and also select lightpaths with acceptable qualityof transmission. The proposed algorithm has better performance improvementwhen compared to typical RWA algorithm.Citations by Others: This paper is cited by Aparicio-Pardo et al. [2011].

6.6 Sengezer et al. 2010

Citation:

Sengezer, N. and Karasan, E. 2010. Static Lightpath Establishment in MultilayerTraffic Engineering Under Physical Layer Impairments. IEEE/OSA Journal ofOptical Communications and Networking 2, 9, 662–677.Problem: The authors state that there is problem in establishing static versions

of lightpath requests which consider physical impairments in the optical layer whichis the backbone of the lower layer in the multilayer architecture which consists ofa Service layer, Electrical layer and Optical layer. The authors state that thereis a need to formulate efficient algorithms in order to satisfy maximum lightpaths(requests) with minimum blocking and bit error rates of the established lightpathsand make efficient use of the bandwidth allocated for transmission of data.Previous Work: The authors refer to previous work by Christodoulopoulos et al.

[2009].Shortcomings of Previous Work: The authors state that the existing method was

not investigated in detail and there was only a basic comparison with other methodswhich do not consider impairments and no further study was done. The authorsstate that there are many other factors which influence a lightpath to satisfy ademand and must be taken into consideration and investigated further.New Idea/Algorithm/Architecture:The authors propose an Integer Linear Pro-

gramming (ILP) formulation for static lightpath establishment which considersphysical layer impairments. The idea behind the ILP formulation is to route asmany lightpath demands by making use of the resources provided. In formulatingthe ILP the authors considered the aggregate effect of the physical impairmentsthrough which low Bit Error Rate is guaranteed which is below the acceptablethreshold. A set of paths for each source-destination are generated and each re-quest is routed through paths chosen and at every level a link capacity constraintis imposed. A physical impairment model is used to provide with upper limit of



impairments a lightpath can handle when it is routed from the source, and the levelat the receiver is monitored using a linear expression which represent the noise leveldue to ASE, Cross-talk and by the signal itself. The authors state that there is adrawback in the ILP formulation as the size of the network increases the solutiontime also increases exponentially, which makes it infeasible for a large size network.To overcome this drawback a polynomial time heuristic approach called ROLE(ReOrdered Lightpath Establishment Algorithm) was formulated taking into con-sideration large size networks the objective of the heuristic is to route as manyrequests as possible taking consideration of the Bit Error Rate constraint and notviolating it. The authors formulated the ROLE heuristic in three phases first isRouting and Wavelength Assignment, rerouting and reordering. In Routing andWavelength Assignment shortest path is selected along with an available wave-length and best combination is looked for a particular lightpath demand. AfterRWA rerouting is performed in order to improve the results of the RWA phase,which has a few lightpaths blocked due to no available wavelength or due to BERthreshold. The algorithm first tries to route the BER blocked demands and thenwavelength blocked demands so that minimum node cross talk is achieved. In caseof reordering the authors state that all the unsatisfied lightpath requests are madeto undergo RWA and rerouting until there are no unsatisfied lightpath demands.

Experiments Conducted: The authors state that in order to evaluate the perfor-mance of the ROLE algorithm proposed, and to make comparison with other ap-proaches as well as to compare with itself based on sorting techniques, a randomlygenerated physical topology each with 20 nodes having 25-40 bidirectional links( topology1, topology2, topology3, topology4) using three different wavelengths4,8,16 and which has the required lightpath demands generated randomly sortedin either shortest lightpath demand, longest lightpath demand was utilized. Ex-periments were also carried out on realistic Deutsche Telekom network topologyin order to make a valid comparison with other proposed procedures. The ILPalgorithm proposed by the author was solved using ILOG CPLEX optimizer whichuses branch and cut method was compared with ROLE for a very small networkand performance was evaluated.

Results: According to the authors experiments on ROLE algorithm with takinginto consideration both shortest demand first and longer demand first on varioustopologies shows percentage of routed lightpaths is from 65 for topology1 to 87 fortopology4.

Claims: The authors claims that the proposed ROLE algorithm performed 14percentage better than the best heuristic algorithm present so far (POLIO-RWA)for the same problem, and is also guaranteed to provide optimum solutions fordifferent sets of demands for different network sizes.Citations by Others: This paper is cited by Sengezer et al.[2012].

6.7 Sengezer 2012

Citation:

Sengezer, N. and Karasan, E. 2012. Multi-layer Virtual Topology Design in Op-tical Networks Under Physical Layer Impairments and Multi-hour Traffic DemandEEE/OSA Journal Journal of Optical Communications and Networking 4 , 2, 78.



Problem:The authors state that there is a need for proper topology design throughwhich we can obtain less or no wavelength conversion at all during transmission ofan optical signal considering physical impairments along its path. There is a needfor design or a approach in which sharing of information between various layersin the wavelength division multiplexing network is taken care of in order to attainblocking-free performance and best utilization of resources available.

Previous Work: The authors refer to work done by Varvarigos et.al [2009], Sen-gezer et.al [2010]

Shortcomings of Previous Work: The authors point out certain shortcomingsfrom the previous approaches referred as well as other techniques which are used tosolve the problem and state that in the previous approaches the solutions obtainedwas not traceable for moderate or large size networks as the whole problem wasdivided into sub problems and there should also be care taken during integration ofsolutions obtained at every phase and if there is a mistake then we would not obtainoptimal solution at all. The authors also state that all the subproblems consideronly one layer and omit other layers present and ignore sharing of information alsoduring its transmission in an optical network and physical layer impairments arealso ignored. A heuristic approach formulated provided solution but it was observedthat excess utilization of wavelengths than used was possible and further does notprovide optimal solution.

New Idea/Algorithm/Architecture: The authors state that their approach is firstof its kind in the area to solve the problem. They introduce, what they claim tobe, a novel multilayer virtual topology design approach in which both wavelengthassignment and physical layer impairments are considered, and which involves in-teraction of control planes of multi-protocol label switching and wavelength divisionmultiplexing layers during topology design and share information using topologyinformation sharing strategy in which a wavelength routing node architecture wasfollowed in physical layer model. The authors also propose two topology designs:a) Tabu search virtual topology design b) Greedy search virtual topology design inorder to solve multilayer virtual topology design. Apart from this a mixed integerlinear programming approach was made to provide a bench mark to evaluate theperformance of other approaches to the multi-layer VTD problem.

Experiments Conducted: The authors state that in order to evaluate the perfor-mance of the VTD algorithms and information strategies with different traffic ratesand patterns changing with time. A wide area network is utilized with NSFNETtopology with 21 links and 14 nodes spread over four different time zones. Apartfrom this inorder to evaluate the topology design algorithms CPLEX 12.1 optimiza-tion problem solver was used.

Results: The authors state that on conducting experiment with 15 percentagetraffic magnitude with 21 links and 14 nodes MILP shows 31.9, GS-VTD shows27.82, TS-VTD with out layer interaction shows 36.7 and TS-VTD with layer in-teraction shows 0.03 percentages of bandwidth blocking and for a whole day timeperiod blocking ratio for a single-hour traffic demand was 3.19 percentage and formulti-hour traffic demand it was 2.40 percentage both having 100 percentage trafficintensity.

Claims: The authors claim that the proposed method solves the multi-layer



logical topology design and LSP routing problems arising in MPLS layers, androuting and wavelength assignment problems in the wdm layers in an integratedapproach that does not necessitate a single common control plane for both layers.The authors also state that MILP formulation approach does not generate goodsolutions and when taking into consideration heuristic approach Tabu search basedalgorithm can reduce the blocking ratio of the GS-VTD algorithm by 35 to 100percentage.

6.8 Sole et al. 2012

Citation:

Sole, J., Subramaniam, S., Careglio, D. and Spadaro, S. 2007. Cross-LayerApproaches for Planning and Operating Impairment-Aware Optical Networks. InProceedings of the IEEE 100, 5, 1118—1129.Problem: The authors state that in a transparent network which transmits the

data in optical mode from end node to end node (thus reducing the expenses ofconverting the signal from optical to electrical and then to optical mode) there arechances of physical impairments accumulating on the signal during transmissionwhich reduces the optical reach of the signal in the network and is also the sourcefor other failures. The authors state that proper planning and operation phasesshould be taken care of in advance in order to avoid such problems, and thereshould also be certain techniques incorporated which take care of both physicaland network layer interaction and which enable proper allocation of the resourcesand requirements so that optimal solution is obtained with maximum allocation ofdemands. In addition to this, the authors state that there should also be certaintechniques which take into consideration integration of measurements of opticallayer performance into algorithms which considers impairments on the lightpathsignals during routing and wavelength assignment for optical networks to be capableof being automated and network configurable.

Previous Work: The authors refer to previous work by Aparicio-Pardo et.al[2011], Varvarigos et.al [2009].Shortcomings of Previous Work:The authors state that change in currently used

fiber capacity from several 10-gigs of data to several 100-gigs or 400-gigs of dataposes technical problems and existing strategies does not apply to such wide capac-ity of fiber and these are yet to be investigatedNew Idea/Algorithm/Architecture: The authors state that a cross layer strategy

was developed which considers physical layer information, to overcome the problemsunder investigation.Experiments Conducted: The authors state that all the strategies developed were

experimented with a 10gigs/sec system and infrastructure.Results: The authors state that there were no raw results obtained as no com-

parison of the different algorithm results were made.Claims: The authors claim that good Impairment Aware RWA algorithms were

developed for protected and unprotected lightpaths with optimal monitor placementtechniques.Citations by Others: There are no explicit implication to this paper by other

researchers in this survey.



6.9 Varvarigos et al 2010

Citation:

Varvarigos, E., Manousakis, K. and Christodoulopoulos, K. 2010. OfflineRouting and Wavelength Assignment in Transparent WDM Networks IEEE/ACMTrans. on Networks 18, 5, 1557—1570.Problem: The authors state that signal quality degrades during transmission

from source to destination due to physical layer impairments and also effects therouting choices made for a particular lightpath as well as other lightpaths whichare transmitted along the same fiber. The authors state that there should becertain optimization techniques which help in providing connections by utilizing aminimum number of wavelengths but also helps in providing quality signal whichis acceptable.Previous Work:The authors do not to any previous work from the list of papers

present which are exactly on topic.Shortcomings of Previous Work: The author states certain short coming in the

previous approach by other authors working on the same problem. The authorsstate that decomposition techniques were followed in previous approaches whichmight not provide an optimal solution for the joint Routing and Wavelength As-signment problem, Apart from this, when taking into consideration interference,existing approaches do not consider the interference due to already establishedlightpaths on the candidate lightpath. Instead they assume a worst case scenariowhere all the lightpaths are active. The authors state that a new optimal tech-nique should be formulated in order overcome existing problems and shortcomingof previous approaches.New Idea/Algorithm/Architecture:The authors state that an initial approach was

developed to solve the Routing and Wavelength Assignment problem without con-sidering impairments based on Linear Programming relaxation formulation in orderto generate integer solutions, then the algorithm is extended to two impairmentaware routing, and a wavelength assignment algorithm was proposed that accountsfor interference within lightpaths. The authors state that one approach takes phys-ical impairments directly into account with the help of utilizing noise related pa-rameters, whereas the other approach takes into effect the physical impairmentsindirectly and formulate a cross layer approach and helps in satisfying maximumnumber of requests (source to destination) with acceptable signal quality until itreaches the destinationExperiments Conducted:The authors state that in order to test the proposed al-

gorithms numerous simulation experiments were carried out in Matlab with randomstatic instances. LINDO was used in order to solve the LP and ILP problems andthe network topology used in their simulations was the generic deutsche Telekomnetwork and the capacity of the wavelength is considered to be 10 Gb/s. In orderto evaluate the feasibility of the lightpaths a QoT tool was used with acceptableQ-factor equal to 15.5 dB. For evaluating the IA- RWA algorithm a link modelwith NRZ-OOK modulation format, 10-gb/s transmission rates and 50 GHz chan-nel spacing with span length of 100km each having SSMF fibers with dispersionparameter of 17 ps/nm/km and attenuation of 25 db/km, was used.Results: The authors state that with traffic matrix load between 0.5 and 3 as



input. The SB-IA-RWA scales well with the load. Considering realistic trafficmatrix of DTnet in the simulation experiments the IA-RWA blocking ratio decreasessignificantly. The SB-IA-RWA can obtain a zero blocking solution, with runningtime about 15 min utilizing around 19700 variables and 22515 constraints for 35wavelengths, about 3.2 percentage and running time around 3 hours utilizing 21338variables and 61278 constraints for P-IA-RWA and 10 percentage with pure RWAand with running time around 30 seconds.

Claims: The authors claim that the approach of solving RWA problem basedon LP-relaxation generates an optimal solution despite absence of integrality con-straints for large instances. The authors further claim that the proposed two IARWA algorithms which are P-IA-RWA and SB-IA-RWA show better performancethan the pure RWA algorithm and moreover SB-IA-RWA shows better performanceminimizing wavelength utilization providing zero-blocking solution with very lessexecution times.

Citations by Others: There are no explicit implication to this paper by otherresearchers in this survey.

6.10 Zhang et al. 2009

Citation:

Zhang, W., Tang, J., Nygard, K., and Wang, C. 2009. REPARE: RegeneratorPlacement and Routing Establishment in Translucent Networks. GLOBECOM2009 - 2009 IEEE Global Telecommunications Conference, 1—7.Problem: The authors state that most of research done in designing and routing

optical signals assumes that the medium in which the data is transferred passeswithout any error in bits. But in fact the situation of non-erroneous transfer of datamay not really occur since there are chances that some physical impairment may beintroduced by the optical fibers or other components utilized to route optical signals.The authors state that such impairments should be taken into consideration whiledesigning and routing optical signal and there should be appropriate techniquesto transfer data with the help of optical signals without any bit errors duringtransmission and look into providing an efficient and effective routing of opticalsignal.

Previous Work: The authors state that there is no previous work referred by theauthors.

Shortcomings of Previous Work:The authors state certain shortcomings in gen-eral taking the over all scenario. The authors state that even though physicalimpairments are considered during transmission of optical signals, there is no as-surance of a particular request being accepted in order to transmit data becausethe signal could possibly generate high bit error rate and requests could be rejected.Hence many optical signal requests can not be accepted because of chances that itcould not keep up the quality of the signal due to impairments, which makes thewhole network inefficient to accept many source and destination requests. Thereforethere should be techniques to convert the inefficient signal to efficient signal eventhough some cost is involved so that we can increase the overall network utilizationup to its maximum capability.New Idea/Algorithm/Architecture: The authors state that during transmission of



an optical signal, impairments are caused by the optical fibers or components usedin the optical network which in turn is responsible for bit errors. To overcome theproblem of bit errors during transmission of optical signals the authors introducea novel Integer Linear Programming (ILP) formulation for small networks anda heuristic approach for a large network which places minimum number of 3Rregenerators (where each R= reshaping, R= retiming, R= reamplification) in anefficient and effective way with optical electrical optical conversion capability toaccommodate all requests.Experiments Conducted: The authors state that in order to compare and evaluate

the performance of the proposed methods the pacific bell network topology was usedto test the network which has 15 nodes and 21 links with 32 wavelengths per fibercapability apart from that CPLEX was also used to solve the ILP formulation togenerate an optimal solution. The simulation was performed on a machine with aIntel Pentium 1.70GHz CPU and 512 MB memory.Results:The authors state that for traffic matrix size upto 70 starting from 50

size, nearly 4 regeneration nodes were required for ILP formulation method and forheuristic method it was varying from five to seven and beyond 70 traffic matrix size,ILP could not generate solution. In case of the heuristic approach it generated afeasible solution as the traffic matrix size increases. The running time performancefor ILP to generate solution was 1000 millisecond whereas for a heuristic to generatea solution it was taking a few milliseconds.Claims: The authors claim that the results obtained from the experiments show

that their heuristic algorithm outperforms the ILP optimal solution for varioustraffic matrices and showed better result and performance than the ILP optimalsolution.

Citations by Others: There are no explicit implication to this paper by otherresearchers in this survey.

7. REFERENCES

REFERENCES

Aparicio-Pardo, R., Klinkowski, M., Garcia-Manrubia, B. , Pavon-Marino, P., and

Careglio, D. 2011. Offline Impairment-Aware RWA and Regenerator Placement in Translu-cent Optical Networks. Journal of Lightwave Technology 29,3, 26–277. (Exactly on topic)

Christodoulopoulos, K., Manousakis, K., and Varvarigos, E., 2009. Considering Physical

Layer Impairments in Offline RWA. Network, IEEE 23, June, 26–33. (Exactly on topic)

Manousakis, K., Christodoulopoulos, K., Kamitsas, E., Tomkos, I., and Varvarigos, E.2009. Offline Impairment-Aware Routing and Wavelength Assignment Algorithms in Translu-

cent WDM Optical Networks. Journal of Lightwave Technology 27, 12, 1866–1877. (Exactly

on topic)

Pavon-Marino, P., Azodolmolky, S., Aparicio-Pardo, R., Garcia-Manrubia, B., Poin-

turier, Y.,Angelou, M., Sole-Pareta, J.,Garcia-Haro, J., and Tomkos, I. 2009. OfflineImpairment Aware RWA Algorithms for Cross-Layer Planning of Optical Networks. Journalof Lightwave Technology 27,12,1763—1775.(Exactly on topic)

Saradhi, C. and Subramaniam, S. 2009. Physical layer impairment aware routing (PLIAR) in

WDM optical networks: issues and challenges. Communications Surveys & Tutorials, IEEE,11,4, 109–130.

Sengezer, N. and Karasan, E. 2010. Static Lightpath Establishment in Multilayer Traffic En-

gineering Under Physical Layer Impairments. IEEE/OSA Journal of Optical Communicationsand Networking 2, 9, 662–677. (Exactly on topic)



Sengezer, N. and Karasan, E. 2010. Static Lightpath Establishment in Multilayer Traffic En-

gineering Under Physical Layer Impairments. IEEE/OSA Journal of Optical Communications

and Networking 2, 9, 662–677. (Exactly on topic)

Sengezer, N. and Karasan, E. 2012. Multi-layer Virtual Topology Design in Optical Networks

Under Physical Layer Impairments and Multi-hour Traffic Demand EEE/OSA Journal Journal

of Optical Communications and Networking 4 , 2, 78. (Exactly on topic)

Sole, J., Subramaniam, S., Careglio, D. and Spadaro, S. 2007. Cross-Layer Approaches forPlanning and Operating Impairment-Aware Optical Networks. In Proceedings of the IEEE 100,

5, 1118—1129. (Exactly on topic)

Varvarigos, E., Manousakis, K. and Christodoulopoulos, K. 2009. Cross layer optimizationof static lightpath demands in transparent WDM optical networks. IEEE Information Theory

Workshop on Networking and Information Theory, 115—119.

Varvarigos, E., Manousakis, K. and Christodoulopoulos, K. 2010. Offline Routing and

Wavelength Assignment in Transparent WDM Networks IEEE/ACM Trans. on Networks 18,5, 1557—1570. (Exactly on topic)

Zhai, Y., Askarian, A., Subramaniam, S.,Pointurier, Y. and Brandt-pearce, M. 2010. Cross-

layer approach to survivable DWDM network design. IEEE/OSA Journal of Optical Commu-nications and Networking, 2,6, 319—331.

Zhang, W., Tang, J., Nygard, K., and Wang, C. 2009. REPARE: Regenerator Placement

and Routing Establishment in Translucent Networks. GLOBECOM 2009 - 2009 IEEE Global

Telecommunications Conference, 1—7.(Exactly on topic)


A Survey on Static Routing Wavelength Assignment...

Documents

Transcript of A Survey on Static Routing Wavelength Assignment...