SDN中基于蚁群优化的网络测量节点选择算法  

Network Measurement Node Selection Algorithm Based on Ant Colony Optimization in SDN

作　　者：叶和元 韩俐 孙士民 YE Heyuan;HAN Li;SUN Shimin(School of Computer Science and Engineering,Tianjin University of Technology,Tianjin 300384,China;School of Computer Science and Technology,Tiangong University,Tianjin 300387,China)

机构地区：[1]天津理工大学计算机科学与工程学院,天津300384 [2]天津工业大学计算机科学与技术学院,天津300387

出　　处：《计算机工程》2022年第5期191-199,共9页Computer Engineering

基　　金：国家自然科学基金(61802281,61702366);天津市自然科学基金(19JCYBJC15800);天津市高等教育委员会科技发展基金(2017KJ090)。

摘　　要：在软件定义网络(SDN)中,当流传输路径信息获取受限时,现有的测量节点选择算法只能基于网络拓扑的中心性指标进行测量节点选择,存在测量精度较低、测量负载不均衡、运行时间长等问题。将SDN网络中测量节点选择问题抽象为最小顶点覆盖模型,提出一种基于蚁群优化的测量节点选择算法ACO-NS。利用复杂网络的度分布理论缩减状态转移过程中的候选集规模,同时设计一种信息素局部增强-全局挥发机制,增大可行解的信息素浓度,提高算法的准确度和收敛度,并且缩短搜索时间。通过OpenFlow消息在线计算测量节点的负载,采用邻域搜索策略对过载节点进行筛选和替换,以降低过载处理的时间。实验结果表明,与ACO算法相比,该算法的准确度和收敛度分别提高56.7和28.2个百分点,且单位时间内的过载处理开销降低79.8个百分点,具有较高的测量精度。In Software Defined Network(SDN),when access to stream transmission path information is limited,existing measurement node selection algorithms can only select nodes based on the central index of the network topology,resulting in low measurement accuracy,an unbalanced measurement load,and a long run time.The measurement node selection problem in SDN network is abstracted as the minimum vertex coverage model,and a measurement node selection algorithm ACO-NS based on Ant Colony Optimization(ACO)is proposed.The degree distribution theory of complex networks is used to reduce the size of candidate sets in the state transition process.Simultaneously,a pheromone local enhancement global volatilization mechanism is designed to increase the pheromone concentration of feasible solutions,improve the accuracy and convergence of the algorithm,and shorten the search time.The load of the measurement node is calculated online through the OpenFlow message,and the Neighborhood Search(NS)strategy is used to filter and replace the overloaded nodes to further reduce the time overhead of overload processing.The experimental results show that compared to the ACO algorithm,the proposed algorithm improves the accuracy and convergence by 56.7 and 28.2 percentage points,respectively,and has high measurement accuracy,and the unit time overhead is reduced by 79.8 percentage points.

关 键 词：网络测量 测量节点选择 蚁群优化 邻域搜索 软件定义网络 

分 类 号：TP393.07[自动化与计算机技术—计算机应用技术]