Research On Clustering Algorithm In Ad Hoc Network And Improvement On The Adaptive On-demand Weighting Algorithm

Mobile Ad Hoc network is multi-time temporary automated system which consists of a group of mobile terminal nodes equipped with wireless reception facility. It's unnecessary for Ad Hoc network to be furnished with central controlling nodes because every node is equal and anti-ruinous. Moreover the node in the Ad Hoc Network simultaneously contains the functions of both host computer and router. On the one side, it carries out the application program for the clients; on the other side, it participates the classified transfer and maintenance of the router based on the routing strategy and routing table. Ad Hoc Network is generally applied for military communication, emergent control, temporary conference, mobile communication and sensor network.The architecture of the Ad Hoc Network is divided into planar structure and hierarchical structure. In the planar structure, all nodes are theoretically equal without bottle-neck nodes so that the net is comparatively strong. However, this structure is less expansionary, being just applied in the smaller size of the Ad Hoc Network. To some extent, the hierarchical structure can solve the problems existing in the planar structure. In the architecture, Ad Hoc Network is classified in clusters which consist of one head cluster and other sub-clusters. The task for the head cluster is comparatively loaded, so it likely becomes the bottle-neck of the network. Therefore, it's essential for the normal function of network to choose the head cluster reasonably. The selection of the head cluster depends on the execution of the clustering algorithm, which performance directly affects the capability of the clustering structure.At the beginning part of the dissertation, the technology of the Ad Hoc Network is introduced briefly, such as the clustering algorithm in the network. And then the advantages and disadvantages of various typical clustering algorithms have been compared in details. In this part, the adaptive on-demand weighting clustering algorithm is mainly discussed. Based on the former theoretical analysis, the adaptive on-demand weight(AOW) algorithm is improved and the network simulator(NS) is also extended for modeling the clustering algorithm. In the final part, the simulation test has been carried out with the expanded NS and after test the improved algorithm and the original one have been compared.It's so general and flexible that we make use of the weight in AOW algorithm to look at different aspects of situation. The application of the AOW algorithm for the military command is concentrated in the paper owing to the typical factor in battlefield– the command weight; AOW algorithm will be improved by adding command weight in order to form the cluster structure, which will be more adaptable for the focus-command in the modern war.The simulation test shows that comparing with the original algorithm, advanced algorithm can reduce the communicational times and greatly improve the performance without obviously increasing the account and communicational expenditure. As the result, the improvement of the AOW clustering algorithm has been proved to be efficient and practical.