Research Of Asynchronous Clustering Algorithm In Wireless Sensor Networks

As a new platform for colleting information, wireless sensor networks have great space of development in many applications, such as military, environmental, medical and home applications. Attracting many researchers'attention, wireless sensor networks have been classified as one of the 21 most influential technologies and the ten techniques changing the world by Business Week and MIT Technology Review.Wireless sensor networks are mainly formed by a large number of sensor nodes though wireless communications, which can perceive information, collect data and process those information and data of the monitoring objects in the networks collaboratively. At last, the information data will be transmitted to the observer. Because it is inconvenient even difficult for users to work in the application environment of wireless sensor networks, so the energy in the nodes can not be replace, and those nodes have to work normally only depending on their own energy. In order to take advantage of the limited energy resources in nodes and improve the life of networks as much as possible, many domestic and international researchers have proposed a series of different routings for wireless sensor networks.In this context, this dissertation analyzes the advantages and disadvantages of the some classical routing protocols abroad. And then studies the good ideas of some domestic clustering algorithm. At last, an asynchronous clustering algorithm-ACA is given, which is dynamic and local.Based on LEACH Protocol, ACA algorithm is the improved algorithm of HEED-SM. The algorithm has the following characteristics:1) Structure of clusters. First of all, the global network will be divided into several clusters randomly in the initialization, and then regions of cluster will be changed with the dynamics changing of the cluster head. In this way, it will solve the problem of close distance between cluster heads, and make the cluster heads located in the center of the clusters. So the ACA can balance the energy consumption of nodes in each cluster and improve utilization of the scant energy.2) Condition of selecting Cluster head. An energy threshold-TE, which is part energy of the most energy in a node during the cluster, will be set up for selecting a new cluster head when a cluster is build. If the energy in the cluster head is less than the energy threshold, the cluster will execute selecting a new head. The energy threshold may be different from clusters in the network. But the energy threshold will drop with the reducing of energy in nodes when the network operates normally, which can balance the energy consumptions during nodes and make the nodes die at the same time as soon as possible.3) Cluster head selection. When a old cluster head discovers its residual energy is less than the energy threshold, which means it need select a new head, it will broadcast the node with the most energy be temporary head, and then the temporary head will judge if some cluster heads existence within a certain radius around itself. If no, the temporary head will broadcast a statement as a new cluster head; otherwise, it will sent a message of repulse, and the old cluster head selects the node with secondary most energy, and so on. So even if some cluster heads may be close to others, but with the emergence of new cluster heads, the distances during heads will be adjusted to a suitable level.Finally, the simulation results of ACA on the MATALAB platform show that the ACA has greater performance than LEACH and HEED-SM protocol on network data traffic, energy utilization and the distribution of cluster structure etc.