中文摘要：

该文针对虚拟网络映射过程中节点与链路映射关联性差、虚拟邻接节点映射距离远和相邻节点链路资源消耗不均衡的问题,提出一种两阶段的基于节点邻近感知与路径综合评估的虚拟网络映射算法（NA-PVNM）。在节点映射阶段,首先对虚拟节点按照资源请求和广度优先搜索算法排序,然后对每个虚拟节点的候选物理节点考虑资源丰富度和拓扑邻近性,建立了节点适应值函数计算最佳映射节点。在链路映射阶段,采用k最短路径算法,对每条候选路径考虑可用带宽、途经物理节点资源最大值和路径跳数,通过路径适应值函数计算最佳路径。仿真结果表明算法降低了链路映射长度,提高了虚拟网络请求接受率和收益开销比。实验分析了位置约束和物理网络拓扑性质对算法性能的影响以及映射过程中物理网络资源占用情况。实验结果表明,在物理资源分布和虚拟网络请求的条件约束下,提高虚拟网络映射成功率的关键在于减少映射时的资源消耗。

英文摘要：

To solve the problems of poor correlation in node mapping and link mapping, wide apart of adjacent virtual node during mapping and imbalance resource consumption of nodes with their adjacent links, a two-stage Virtual Network Mapping algorithm is proposed based on Node Adjacent-awareness and Path comprehensive evaluation （NA-PVNM）. In the stage of node mapping, firstly, virtual nodes are ranked according to resources request and breadth-first search, secondly, a node fitness function is set to find the best node in candidates of a virtual node, which takes resource richness and topology connection feature into account. In the stage of link mapping, a path fitness function is set to find the best path in candidates, which takes available bandwidth, node resource and hops of path into account. Simulation results show that the path distances of virtual links are reduced, the acceptance ratio and revenue/cost ratio of virtual networks are improved using the proposed NA-PVNM algorithm. The influence of location constraint and substrate topology feature on algorithm performance, and the resource occupancy of substrate network during mapping are analyzed by experiments. Experimental results show that, under the constraint of physical resource distribution and virtual network requests, the critical factor of improving success rate is to reduce resource consumption during mapping.