Research Of Dynamic Entropy Combination Weighted Clustering Algorithm For Ad Hoc Networks

Ad hoc network is a self-organizing network of mobile devices connected by wirelesslinks. Clustering algorithm is an effective technique for Ad hoc cyber source management,which can administrative overhead, manage mobile node and channel access control moreeasily, and also can improve the efficiency in networks resource using. In addition, thestability of the cluster structure will have direct effect on throughput, energy consumption,delay and packet loss rate.At present, weight-based clustering has become the mainstream clustering algorithm ofAd hoc networks for it excellent performance. The design idea of the algorithm is todetermine the combination weights according to different factors and its importance. However,most of these clustering algorithms only focused on the calculation method of the factors,have not in-depth studied on factors weights allocation and how to clustering effectively aftergetting weights. Due to weights of different factors in different proportion corresponding todifferent application requirements, how to design the factors weight calculation method is aprincipal problem faced in clustering algorithm. Besides, Ad hoc network is a dynamicprocess, the relative importance of each factor always in a dynamic process of changing.Consequently, it’s necessary for completing a reasonable clustering to consider the dynamiccharacteristic of the nodes when calculating the weight of each factor in Ad hoc clusteringalgorithm.The thesis has a research on the dynamic entropy combination weighted clusteringalgorithm of Ad hoc network. And the structure of thesis is organized as follows: Thedefinition of Ad hoc network, background, application and key technologies are introduced inchapter1; Ad hoc network topology and application are described in chapter2; Chapter3presents a dynamic entropy combination weighted clustering algorithm of Ad hoc networks,and analysis of the performance of the algorithm; Simulation of the optimization algorithmare introduced chapter4; The conclusion is in the end of this thesis, which summarizes thewhole content and looks into the future.Chapter3and4are the focus research for the thesis. Chapter3proposes one kind ofnodes’ dynamic entropy calculation method improved based on traditional entropy method,which incorporates the stability of the networks to the nodes weight calculation. Dynamicentropy not only takes into account the node factors difference at a certain moment, but alsothe boundary value deviations of several messages records, using the optimization algorithmto determine the weight. Combine the weight determined by dynamic entropy method andobjective evaluation method to get combination weight and accomplish clustering. In addition,this chapter also puts forward a new Monte Carlo optimization for clusters maintenance to avoid the frequent clusters replacement phenomenon caused by instability of networksenvironment. Chapter4carries on performance simulation for proposed algorithm, simulationresults demonstrate the superior performance of the improved dynamic entropy combinationweighted clustering algorithm in terms of the networks stability (average number of clusterheads, dominant update numbers, etc).