Research On Clustering Algorithm Of Wireless Sensor Networks

As long as the Internet of Things was officially listed as one of the five emerging strategic industriesof the country, the wireless sensor networks has become one of the hot research at home and abroad. WSNs(Wireless Sensor Networks) consist of large numbers of distributed nodes that are generally randomlydeployed in a region of interest. In typical, each node is equipped with one or more sensors, embeddedprocessors, low-power radios, memory, power supply and sometimes mobility module, location moduleand actuator as well. While these nodes are tightly constrained in terms of battery power and randomdeployment in target areas; make it almost impossible to recharge/replace the dead battery. In order toprolong the network lifetime, energy efficiency protocols should be designed for the WSNs. Efficientlyorganizing sensor nodes into clusters is useful in reducing energy consumption.Firstly, this paper introduces the concept and characteristics of the wireless sensor networks. Secondly,we analysis several classical clustering protocols, and summarize their advantages and disadvantages. Wealso research the performances on the energy efficiency, the stability of the clusters, the scalability, thepropagation delay, the load balance and the complexity of the algorithm. Thirdly, this paper proposes aprotocol, ENCM (Expected Number of Cluster Members clustering algorithm), that designed for one-hopwireless sensor networks in which plain sensor nodes communicate directly with the cluster heads. Allnodes are commonly resource constrained and have limited amount of energy. In wireless sensor networks,BS (Base Station) is generally set up away from the target area. When cluster heads transmit data directly tothe BS, the cluster heads further away from the BS consume more energy related to distance and tend to diefaster, leaving areas of the network uncovered and resulting in shortening the lifetime of the network.Aiming to prolong the network lifetime, we use ENCM to group the cluster members of different clusterheads and push the energy consumption of every cluster head to achieve the average value which iscalculated in this paper. Simulation result and comparison with previous protocols prove that our newalgorithm is able to extend the network lifetime observably and moderately reduce the variance of energyconsumption by the cluster heads.