Research Of Cross-layer Optimization Algorithm In Virtual Network

With the communication network giganticism, the heterogeneous networkdiversification and the explosive growth of the multimedia traffics, virtualizationnetwork optimization technology is a vital way to handle the conflict between thetraditional network architecture and traffics performance optimization. Networkvirtualization diversifies the communication network architecture into separate virtualnetworks, which are independent of each other on the logic. The virtual networks canindependently control the traffics so that the more quickly, conveniently traffics wereprovided to users. A fundamental problem in the virtual networks(VNs) is aperformance optimization. Effective performance optimization algorithm can improvethe Quality of Service(QoS) and reduce the cost. Accordingly, it is significant for us toresearch the effective optimization algorithms in virtual networks. Meanwhile,significant performance gains can be achieved by cross—layer approaches in the virtualnetworks. In this dissertation, under the guidance of communication system, networktheory and convex optimization, the cross-layer optimization techniques in the virtualnetworks are intensively studied. The contents of this work are listed as follows：First of all, the evolution of communication network architecture and thedevelopment of network optimization method are introduced. The concepts andoverviews of virtual network are discussed. Moreover, the basic mathematical modeland solutions of optimization problems in communication network are analyzed,including linear optimization, nonlinear optimization and intelligent optimization.Second, the cross-layer design and convex optimization techniques incommunication network are studied in particular. Aiming to solve the performanceoptimization problems in wireless mesh networks, the advantages between layeredoptimization and cross-layer optimization are compared through the analysis andsimulation. The simulation results show that the cross-layer design is used in wirelessmesh network which can be obtained better network throughput. On the other hand, theoptimization problems of the virtual network are investigated with the convexoptimization theory, and a global cross-layer optimization method based on the convexoptimization technique is proposed, which is been modeling analyzed.Finally, in the virtual network based on the wireless mesh network, the cross-layeroptimization problems are investigated according to the above scheme. In two virtual wireless mesh network classified by traffics, we consider the problem of jointoptimization of source coding, power control, ARQ control, and delay partitioningfunctionalities. Our studied problem is to maximize the streams quality under strictend-to-end delay constraints through adjusting the source coding rate, end-to-end delaydistribution, and each node’s transmit power. Then, the performances of application,physical, MAC, and network layers are modeled by some classical models underreasonable assumptions. We formulate the studied problem as a mathematicaloptimization problem, which is proved to be a nonlinear but log-convex one. A globalsolution based on the interior point algorithm is proposed. And the solution is proved toconverge to the global optimum of the above problem. Meanwhile, the algorithmanalysis and simulation results show that the proposed method can achieve betternetwork performance in a virtualized wireless mesh network.