Research On Topology Control Of Wireless Sensor Networks Based On Improved K-means Algorithm

The wireless sensor network is a new type of information acquisition and processing technology that integrates sensors,networks,wireless communications,and embedded computing.Nodes with small size and lightweight make the network flexibly applicable to the perception and collection of complex environmental information.In addition,network with denseness and selforganization nodes can achieve real-time monitoring of objects.With the development of science and technology,wireless sensor networks are showing their huge application potential in scientific research,military,national defense and other fields,which has drawn great attention from researchers.But the nodes of the network have their limited,such as limited computing capability,limited energy,non-rechargeable,and high recovery cost,thus,it is very important to know how to achieve efficient utilization of node energy and optimal control of the network.This thesis starts with the topology control of the network and trying to make the energy consumption of the network more efficient and to achieve the purpose of improving the network's lifetime.Firstly,base on the K-means algorithm and the knowing of the shortcomings of the LEACH(Low Energy Adaptive Clustering Hierarchy)algorithm propose an improved control algorithm.The new control algorithm is composed of a clustering algorithm,node's path planning algorithm named PSR(Positive Semicircle Rule),and cluster head selection algorithm named MCH(Mark of Cluster Head).Secondly,based on the new improved control algorithm,and used the sleep scheduling to control the network with high node-density,and propose a new sleep scheduling algorithm named IRS(Improve Random Sleeping)and a new cluster head selection algorithm named SCH(Super Cluster Head)algorithm which is superior to the MCH,with these two new algorithm to optimize the network sleep scheduling.The main specific works of this thesis are as follows:1.Inspired by the partition clustering algorithm,propose a new clustering control algorithm which different from the LEACH algorithm.Firstly,the new clustering control algorithm derives the network energy consumption formula to find the ideal value of ‘K';secondly,divided the network into K clusters;then are selected the cluster head nodes in the cluster.With this new clustering control algorithm,the actual cluster head nodes can be evenly distributed in the network.2.Inspired by the Floyd algorithm and Dijkstra algorithm,propose a PSR algorithm based on the maximum weight.And uses this algorithm to find out the best path of the node to the cluster head so that the energy consumption of the node can be reduced.And then,based on the PSR.propose an even better algorithm named TCM(Triangle-C Minimum)algorithm.Experimental data shows that the TCM algorithm can find out a more suitable path of the node to cluster head than the PSR,and can reduce the number of redundant nodes in the path.3.The K-means algorithm uses the cluster centroid as the virtual cluster head.Based on this,propose a cluster head selection algorithm named MCH to select the actual cluster head node in the wireless sensor network.The experimental data show that the MCH algorithm reduces the network's energy consumption and extends the network's lifetime.4.Although the MCH has its superiority in network control,the algorithm itself miss considerate some network cluster head selection factors.Therefore,propose a new cluster head selection algorithm named SCH based on the MCH algorithm.SCH considerate more cluster head selection factors,such as the location of the base station.Experimental results show that the SCH algorithm has better performance over the MCH algorithm,and it has a more positive impact on reducing energy consumption and extend the lifetime of the network.5.In order to reduce the energy consumption caused by sending the redundant data and caused by the idle node that monitoring the network,propose a layer sleep scheduling algorithm named IRS base on the random sleep scheduling algorithms.the advantages of the IRS algorithm.The comparison of experimental data shows that the IRS algorithm has a better performance than random sleep scheduling algorithms.The IRS algorithm can have a better sleep scheduling control on the nodes in the network.