Dynamic Allocation of clustering Technique in Ad Hoc Wireless Network
| Autor: | Chiung-Ying Wang, 王瓊英 |
|---|---|
| Rok vydání: | 2001 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 89 An Ad Hoc network is a temporary network formed by a collection of mobile nodes without the aid of any centralized coordinator. The Ad Hoc wireless network is adaptable to the highly dynamic topology resulted from the mobility; hence it is hard to predict and manage. Clustering is a technique to divide nodes that randomly moving in the network into several groups and also be used for controlling the spatial reuse of the shared channels. Moreover, it can reduce communication overhead and is easy to manage network resource in Ad Hoc wireless network. Many techniques have been proposed in previous researches. For example, Lowest-ID cluster algorithm and highest-connectivity algorithm are used to elect cluster-heads. Distributed and mobility-adaptive clustering (DMAC) is proposed for dealing with cluster nodes in highly dynamic network topology. In addition, the techniques of clustering based on power control are used to promote the system performance and elaborate the power economy. Other parameters, such as cluster radius, cluster dismissed distance and hop distance are also proposed as alternatives of traditional power transmission range to evaluate system performance. In this paper, we propose a novel clustering technique, which is called Dynamic Allocation of Clustering Algorithm (DACA). The DACA algorithm is based on the moving displacement and direction in a certain amount of time to re-cluster or handoff nodes to the proper cluster dynamically, especially for the overlapping nodes between clusters. In addition, we also consider the optimal overlap percentage between clusters to reduce communication overhead. If the overlap percentage is larger than threshold, the overlapping nodes must be re-clustered. Therefore, DACA can enhance topology stability and mobility management, and it is power saving as well if the transmission power level is controlled. Comparing the existing algorithms with the proposed algorithms by evaluating system parameters, such as lowest-id cluster algorithm, highest-connectivity cluster algorithm, multi-hop with lowest-id clustering technique, and multi-hop with highest-connectivity clustering technique, DACA with lowest-id clustering technique, and DACA with highest-connectivity clustering technique, respectively. From the experiment results, it can demonstrate the feasibility and practicability characteristics of the proposed method. We conclude that our method has better performance especially in highly dynamic network, such as (i) reducing the dropping probability of cluster-head, (ii) increasing stabilities of the cluster structure in terms of the probability of nodes that change their roles, the probability of nodes that change clusters (handoff), (iii) providing simple power saving mechanism of cluster structure. Moreover, we also show the proper overlap percentage threshold with different various mobility speeds by computer simulation. Currently, there are many routing protocols are developed based on the existing cluster techniques. In the future, we also can further investigate the related works based on the well-developed DACA techniques. |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|