Research On Virtual Network Embedding Algorithm For Optimizing Resource Allocation Over Elastic Optical Networks

Due to the rapid development of diversified emerging applications and the continuous increase of users' requirements,backbone networks with limited bandwidth resources are suffering great challenges.Although the elastic optical networks use much finer granularity to allocate the spectrum resources,which can improve the transmission capacity of the network to some extent,however,its strict constraints of spectrum continuity and contiguity limit the further increase of spectrum resource utilization.Meanwhile,the explosive growth of data traffic and the continuous expansion of the network makes the energy consumption increasingly serious,which hinders the green and sustainable development of the Internet.Network virtualization technology allows multiple separate and heterogeneous virtual networks to share the common physical infrastructure resources,which can significantly improve the resource utilization of the underlying network and provide technical support for energy-efficient network deployment.Therefore,this thesis introduce the network virtualization technology over the elastic optical networks,and studies the resource allocation problem for virtual network embedding over the elastic optical networks,so as to improve the spectrum resource utilization of the elastic optical networks and reduce the energy consumption of communication equipments in the network.To address the spectrum fragmentation and the problem that virtual links may be embedded to the longer physical lightpaths,a coordinated virtual optical network embedding algorithm based on resources availability-aware(RA-CVONE)is proposed in the third chapter.Firstly,considering the embedding state of adjacent virtual nodes,the node importance metric is designed and a coordinated embedding approach is adopted,which can shorten the length of the embedding lightpaths.Moreover,we jointly evaluate the resources availability of a physical lightpath and its node,then select the candidate lightpaths with less spectrum fragmentations and abundant resources for a virtual link,so as to reduce the failure probability of virtual link embedding.At last,a path selection strategy is proposed to determine the embedding lightpath of which the vacant spectrum resources can be effectively used.Simulation results show that the RA-CVONE algorithm can select shorter physical lightpaths to embed the virtual links,and can effectively improve the spectrum resources utilization of elastic optical networks.To address the energy consumption problem and imbalance resources allocation in virtual optical network embedding over elastic optical networks,a joint Power Consumption and Load Balancing algorithm for Virtual Optical Network Embedding(PC-LB-VONE)is proposed in the fourth chapter.During virtual node embedding,we select the physical node with shorter length of its adjacent links,smaller difference between its computing resources and spectrum resources of its adjacent links,so as to reduce the energy consumption and balance the network load.During virtual link embedding,considering the spectrum coherence and average traffic load of a lightpath,a path resource matching formula is designed and a single-path embedding strategy based on maximal clique is proposed to jointly optimize the power consumption and the traffic load.When single-path embedding fails,to further improve the acceptance probability of virtual optical networks requests,a multi-path embedding strategy based on average link load will be executed.Simulation results show that the proposed PC-LB-VONE algorithm can decrease the energy consumption and balance the network load.