关于跨域虚拟网络的优化狼群映射研究仿真

Research and Simulation of Optimal Wolf Swarm Mapping for Cross-Domain Virtual Networks

虚拟网络映射的目的是将网络底层物理资源,以高可用低开销的方式配置到虚拟网络中,进而提高物理网络的业务扩展性能。针对分布式跨域带来的网络资源异构特性,现有映射算法往往存在节点或链路负载不均衡,资源开销过大,以及报文抖动等问题,提出了优化狼群的跨域虚拟网络映射算法。由于跨域虚拟网络映射过程中,额外的资源开销主要来源于域间,因此算法将映射处理分为域内与域间两部分进行独立分析。对于域内映射只引入元胞结构,增强单目标优化处理性能,将节点采用二进制表示,并设定每一位作为一个元胞,建立节点元胞模型,通过更新元胞与近邻得到域内节点与链路资源的最优配置;对于域间映射,则在元胞基础上,引入优化狼群算法,元胞结构提高搜索的分布能力,优化狼群提高全局寻优性能,利用探狼四处游走,在元胞向量中搜索解,同时得到头狼信息,头狼产生召唤行为通知猛狼目标解的信息,从而利用分工协作实现节点与链路最优解的搜索。仿真结果表明,提出的优化狼群网络映射算法能够有效应对跨域异构资源问题,均衡节点和链路的负载,显著降低网络映射开销和网络映射执行时间。

The purpose of virtual network mapping is to configure the underlying physical device resources of the network into a virtual network in a way of high availability and low overhead,thus to.improve the service scalability of physical network.Specific to the heterogeneous characteristics of network resources brought by distributed cross-domain,existing mapping algorithms often have problems such as unbalanced load of nodes or links,excessive resource overhead,and message jitter.Therefore,a cross-domain virtual network mapping algorithm is proposed to optimize wolves.Because in the process of cross-domain virtual network mapping,extra resource overhead mainly comes from inter-domain,the algorithm divides the mapping processing into two parts,intra-domain and inter-domain,for independent analysis.For intra-domain mapping,only cell structure was introduced,enhancing the performance of single-objective optimization processing.The nodes were represented in binary,each bit was set as a cell,and the node cell model was established.The optimal allocation of node and link resources in the domain was obtained by updating cells and neighbors.For inter-domain mapping,on the basis of cells,the optimal wolf swarm algorithm was introduced.Cellular structure improves the distribution ability of search,optimizes wolves and improves global search performance.Using wolf detection to wander around and search for solutions in cell vectors,the information of wolf head can be obtained.The wolf produced calls to inform the target of the wolf.Finally,the optimal solution of the most nodes and links was searched by means of division of labor and cooperation.The simulation results show that the proposed optimization wolf swarm network mapping algorithm can effectively deal with cross-domain heterogeneous resources and balance the load of nodes and links.The overhead of network mapping and the execution time of network mapping are significantly reduced.