Virtual Optical Network Mapping Algorithms Based On Load Balancing For Elastic Optical Networks

The rapid development of network technologies puts forward the higher requirements on existing network infrastructures.Elastic optical networks(EONs)can allocate suitable spectrum resources accommodating to the requested rate of users,which can increase spectrum efficiency and flexibility of resources allocation with less change to existing network architectures.Therefore,EONs is expected to be an effective solution to the problem of the low resources utilization ratio in traditional wavelength division multiplexing(WDM)networks.Different services have their own manners of network resource usages and the requirement for quality of service,which makes it difficult for the existing optical network architecture to meet the dynamic demands of various network services,thus leading an urgent problem in current research area to deploy a set of physical network infrastructures.Network function virtualization(NFV)can insulate the difference between the physical network resources and service requirements needed by different types of users,this cost-effective technology makes it possible for network operators to realize overall planning to meet the needs of different users,and achieves optimal allocation of network resources.This thesis studies on the problem of virtual optical networks(VONs)mapping under EONs.By considering the information of physical network resources usage,mapping strategies from VONs to EONs are proposed to achieve a better performance in network load balancing and resources utilization ratio.The first chapter introduces the development status of traditional optical networks and the evolutionary trend in the future,outlines the research status of VONs mapping problem,and gives difficulties of VONs mapping in EONs and shortages of current researches.The second chapter introduces the mapping principle of VONs and sorts of mapping algorithms.The definition of the two-layer network model that is composed of the underlying physical network and the upper virtual network is given.The comparison of advantages and disadvantages of existing various mapping strategies is made.In addition,this chapter gives basic principles of greedy heuristic algorithms,and analyses strengths and weaknesses of heuristic algorithms used in optimization problems.Aiming at efficiently mapping multiple VONs on the same substrate network,the third chapter builds up an integer linear programming(ILP)model of VONs mapping in EONs,and proposes a load balancing algorithm that is based on key link and resources contribution degree(RCD)for VONs mapping(KLRC).In this proposed algorithm,virtual nodes mapping is completed according to the proposed concept of RCD,which comprehensively considers computing resources of a node,its node position in the network and available bandwidth in links that connect to such a node.While link length,occupied link bandwidth and key link parameters are considered to realize virtual links mapping,thus the appearance of bottle-neck links in a network could be avoided.Simulation results show that,compared with the existing improved-closeness(IC)algorithm under the same environment,the proposed KLRC algorithm occupies less physical network spectrum resources and improves network load balancing.In the process of solving the VONs mapping problem in EONs with heuristic algorithms,the dealing sequence of user's requests and the parameters setting in the algorithm will have much influence on optimization results.Besides,heuristic algorithms solve the mapping problem according to only the current status and the empirical methods,which lacks the guidance on optimization process and may make heuristic algorithms instable.In order to solve such a problem,the fourth chapter proposes a VONs Mapping algorithm that is based on Decomposition and Feedback Adjustment and referred to as DFAVM.By using the concept of decomposition in dealing with VONs requests,the algorithm can determine a more suitable mapping order for each virtual node and each virtual link,and decrease the negative effect caused by VONs mapping order.Through feedback adjustment,the proposed algorithm can get some mapping schemes that can achieve relatively better optimization results among the entire mapping schemes set during mapping the current virtual request,and then calculate the change of the global optimization objective for each mapping scheme,thus choosing the best mapping scheme for the request from the global perspective.Finally,by comparing the simulation results with those existing algorithms,the reliability of the proposed algorithm is verified.The fifth chapter summarizes the work of this thesis,and gives the future research from three aspects: the underlying physical network architecture,the mapping optimization objectives of VONs and the solving algorithms.