基于学习自动机理论与稳定性控制的自适应移动无线Ad Hoc网络分簇策略  

An Adaptive Clustering Strategy for MANET Based on Learning Automata Theory and Stability Control

作　　者：郝圣 张沪寅[1] 宋梦凯 HAO Sheng ;ZHANG Hu-Yin ;SONG Meng-Kai(School of Computer Science,Wuhan University,Wuhan 430079)

机构地区：[1]武汉大学计算机学院,武汉430079

出　　处：《计算机学报》2018年第9期2089-2105,共17页Chinese Journal of Computers

基　　金：国家自然科学基金(61772386);广东省省级科技计划项目(2015B010131007)资助~~

摘　　要：移动无线Ad Hoc网络(MANET)是一种自组织、自配置的多跳无线网络.它不依赖于预先存在的基础通信设施或中心管理方式.分簇策略已被证明是一种能够模仿固定通信设施并提高网络可扩展性的有效途径.它能够将网络划分成若干子网并且被广泛地用于网络管理、资源管理、层次路由设计、服务质量改善与网络安全检测.在移动无线Ad Hoc网络中,频繁的节点移动、不佳的节点分布会减少簇的生存时间、降低通信质量并增加通信开销,而这些问题势必会降低簇的稳定性.此外,如何在动态环境下有效地调整分簇结构也是我们需要重点考虑的问题.一种高效的分簇策略应该是自适应的并且能够根据当前的网络环境与节点状态预测自身的分簇行为变化.针对上述问题,该文提出了基于学习自动机理论与稳定性控制的自适应MANET分簇策略(LASCA).文中首先推导出簇的期望生存时间模型与簇的可靠性通信度量模型.在此基础上,该文设计了簇首的稳定性度量模型进行簇首选择.该模型能够保证所选择的簇首节点构成的簇具有较好的生存时间且节点分布产生较小的通信开销,同时具有较高的通信可靠性,即提高了分簇的稳定性.针对以往工作并未考虑如何在动态分簇重构过程中降低簇首选择开销的情况,该文随后利用学习自动机理论构建了分簇行为认知模型,给出了节点簇首选择行为与概率函数的映射关系,并通过感知网络环境反馈对概率函数进行更新,从而有效地调整动态环境下的分簇结构,减少了不必要的簇首选择计算开销.实验结果表明,该文提出的分簇策略在稳定性指标方面有很好的表现,有效地降低了簇首节点与成员节点的更新次数,并在一定程度上降低了通信开销与簇首选择的计算开销.其中,在最大移动速度为10m/s的Random waypoint移动模型中,WCA、FCA、TVCA的簇首节点累积更�Mobile wireless Ad Hoc network(MANET)is a self organizing and self configuring multi hop wireless network,which does not rely on any pre-existing communication infrastructures or centralized management.Clustering has been proven to be a promising approach(through dividing the network into several sub networks),which can imitate the operation of the fixed communication infrastructure and improve the scalability of mobile wireless Ad Hoc network.It has been widely used in efficient network management,improving resource management,hierarchical routing protocol design,improving quality of network service and detecting the mobile wireless Ad Hoc network security.In mobile wireless Ad Hoc network,frequent node motion,bad nodes distribution can reduce the survival time of clusters,degrade communication quality and increase the communication overhead.All of these problems will certainly decrease the stability of clusters.Besides that,how to effectively adjust the selection of cluster heads in this dynamic network environment should be taken into account.A high-efficiency clustering strategy should be adaptive and can predict the variation of clustering behavior on the basis of current network environment and node state.Focusing on these problems,we put forward an adaptive clustering strategy for MANET based on learning automata(LA)theory and stability control(LASCA).Firstly,we derive the cluster’s expected survival time model and reliable communication metric model.Based on these two models,we design the cluster’s stability metric model as the measurable indicator of cluster head selection.This model can guarantee the selected clusters have relatively high survival time and good nodes distribution,which lead lower communication cost and better communication reliability,i.e.,enhancing the stability of clustering.Aiming at the fact that previous work do not consider how to reduce overhead of cluster head selection during the dynamic clustering re-construction process,we construct a clustering behavior cognitive model b

关 键 词：移动无线Ad HOC网络 分簇 学习自动机理论 稳定性度量模型 分簇行为认知模型 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]