| With the development of computer network deeply into people's daily lives and works, people have reached a dependence on it more and more. Mobile ad hoc wireless networks or may called MANET, roled as an important complement of traditional networks, have attracted much attention of researches in recent years. It has acted a insubstitute part in many applications. What this paper have done was improve Automatic On-demand Weighted Clustering Algorithm (AOW), mainly concentrate on the improvement of topology stability and load balance of the whole networks, and to develop a emulator to varify the improve algorithm.In the first part of this paper, concept and character and key technology and applications of mobile ad hoc networks was introduced and researches on it from civil and abroad was analysised. After the discriptions of demands and criterias of clustering algorithms in ad hoc networks, we analysised some classic clustering algorithms by comparison. In the second part of this paper we defined the concept of mobile-holding-time of nodes in ad hoc networks and proposed a general method for calculating by which we can get a better predicate of nodes mobility. Base on it we proposed a new clustering algorithm called IAOW to improve the AOW clustering algorithm and the astringency of IAOW is proved. In the last part of this paper we designed an emulator platform for running of clustering algorithms using an mechanism called disperse-time-advance, and users can add their new clustering algorithms into it easily. The platform was developed by Delphi7.0 under Windows XP . We mainly discussed the theory and the framework and data structure of the platform as well as expand interface of it. We running IAOW on it compare with AOW , mainly concentrate on the cluster refreshment under the gradually enhencement of nodes' mobility, and found that the performance of IAOW is better than the latter in the large scale and high mobility of nodes networks environment. |
PDF Full Text Request