Virtual Network Embedding Problem In Optical Networks

Faced with such high-speed development of cloud computing, the Internet business diversification and the explosive growth of user demands, network resources will be particularly tight. To alleviate network resources such intense pressure, network virtualization technology is introduced. As a basic problem of network virtualization technology, virtual network mapping problem solves resource allocation problem of this technology. Because the different underlying network has its own features, virtual network mapping in different physical networks needs to combine different constraints.Elastic Optical Network can allocates spectrum resources flexibly to build an all-optical channel according to the transmission rate and modulation formats.Therefore the researches on virtual network embedding problem in elastic optical networks has a certain significance.This thesis mainly focus on two problems in elastic optical network.One is modulation mode selected virtual network mapping problem,the other is energy consumption awareness virtual network mapping problem.In the modulation mode selected virtual network mapping problem, this thesis designs a ILP model based on the continuity and non-overlapping of spectrum grids.The model minimizes the mapping resource cost as the objective function, and uses routing constraints, resource constraints, the choice of modulation mode and distance constraints for the constraints of the model.And according to the result of the model, the thesis designs a two-stage mapping heuristic algorithm. In virtual node mapping stage of the heuristic algorithm, this thesis designs Node Integrated Selection Factor(NISF) and load balance as the core content. The NISF of node takes into consideration the factors such as node topology and link distance. In virtual link mapping phase, firstly the thesis searches K shorest paths between all node pairs in the physical network, and then traverses the K paths in turn.Then during the traversal of physical paths, the algorithm computes path length, chooses a suitable modulation mode, calculates the actual grids of the virtual link, and searches the public continuous avaiable grids in the physical path. Finally, the thesis compares the results of model and algorithm in performances that the algorithm is superior to the model in load balancing and solving time and inferior to the model in mapping cost and blocking probability.What’s more, due to the emergence of the global energy crisis, this leads to how to reduce the network energy consumption and the cost of network operations as an urgent problem.Therefore, this thesis studies the problem of the energy consumption awared virtual network mapping in elastic optical networks, Through analyzing the energy consumption characteristics of nodes and links in elastic optical networks and based on the energy consumption calculation formula, a ILP model is designed. In the model, the thesis minimizes energy consumption as the objective function, and takes advantage of routing constraint, node and link activated constraints for the constraints. According to the characteristics of mapping, this thesis also designs an algorithm. In the virtual node mapping stage of the algorithm, the thesis minimizes energy consumption in nodes as the goal based on making node embedding success as much as possible. In virtual link mapping phase, this article minimizes energy consumption in links as the goal. The algorithm firstly searches the embedding results in on-status physical network, then in global physical network if failure in on-status physical network. Finally, the thesis compares the results of the model and algorithm in performances that the algorithm is superior to the model in load balancing and solving time and inferior to the model in mapping energy consumption and blocking probability.Finally, this thesis summarizes the full text of the research work, and is prospected for the future work.