Research On The Clustering Algorithm And Its Application In Ad Hoc Network

Ad Hoc networks have attracted more and more attentions recently due to itsflexibility and self-organization, which, however, brings big challenges to manage thiskind of dynamic network. Clustering algorithm has been proposed to improve the networkperformance and reduce the system cost. Aiming to propose new clustering methods,existing algorithms and their applications are critically reviewed in the thesis according totheir characteristics and performance. Two novel clustering algorithms are presented,which are different from each other in application scenarios. The first one is based on therelationship of clique and independent set, which is able to generate clusterhead andgateway. With respect to message and computation complexity, the algorithm outperformsthe existing methods, especially in emergent cases. Another approach focuses on theproblem of clustering algorithm without any considerations of security. A secureclustering algorithm based on trust evaluation scheme is proposed and tested in asimulation environment. Its security is higher than other methods and can be used insecurity requirements occasions.Security has been a key issue of Ad Hoc research. The approaches of clusteringbased key management and intrusion detection are presented, which aim to improve AdHoc network security. The key management method has been designed by VC++6.0 andis better than other protocols. It seems that the intrusion detection method outperformsother strategies in network architecture. The applications in security of clusteringalgorithm show that the clustering strategies are with good performance.