Research On Node Clustering Algorithm And Scheduling Strategy For Wireless Sensor Network

A wireless sensor network (WSN) is composed of a large number of densely deployed sensor nodes, which is characterized by low-cost, high-robustness, rapid deployment, self-organization. WSN has been called as a ubiquitous technology. It has wide applications in military, health, and smart home. As a new sensing technology, WSN has brought many challenging research topics for researchers. Clustering is an important mean to design routing protocol and network topology. Node scheduling strategy can allocate nodes’states properly and decrease the density of nodes to prolong the life of the network. So, the research on node cluster and scheduling problem are of great significance for the basic theory and the application technologies of WSN.In this thesis, the necessity and key techniques of node clustering and scheduling technology of WSN is analyzed systematically, and make a thorough study in node cluster and scheduling technology target-tracking oriented. Then, energy efficient node clustering and scheduling algorithms are proposed which can efficiently save nodes’energy and prolong network life under precise and total information collection.Aiming at the construction and reconstruction of dynamic cluster for target tracking with wireless sensor network, a dynamic cluster algorithm based on prediction is proposed in this thesis, in which construction and reconstruction of dynamic cluster depends on prediction of nodes’location. Selection of cluster head is achieved by weighted algorithm considering residual energy and geographic position or other parameters. Accuracy of target tracking is improved by cooperation in nodes and the energy consumption in target tracking is described and analyzed. Simulation results show that the proposed algorithm can efficiently save nodes’ energy and prolong network life under accuracy of target tracking compared with the current methods.Target tracking algorithms based on dynamic cluster that has been proposed provide the clusters are not intersecting, that is, there is no common node among the clusters. If a cluster head doesn’t work, the information in the cluster will be lost. A dynamic cluster with maximum entropy is proposed, which topology of overlapped cluster is employed for the presence of common nodes in adjacent clusters. If a cluster’s energy is not enough to realize remote communication, the common node can be used as gateway. While a cluster doesn’t work, the member nodes in the cluster will send information to other cluster by common node to prevent from losing date. Simulation results show that the proposed algorithm can make the cluster head distribute evenly and the number of dead nodes is decreased, so the energy of nodes can be distributed evenly to prolong the life of the network.Aiming at changing the nodes’state and energy consumption, a sensor scheduling algorithm based on prediction. The proposed algorithm employs Gaussian particle filter to predict the node’s position. The nodes in sleeping state in the target occurring area next time can be awaked immediately. Considering the large energy consumption of cluster head, PSO (Particle Swarm Optimization) is adopted to choose cluster head. The fitness function is built by node’s energy consumption model. Optimal cluster head can be achieved by PSO in that the energy consumption between cluster head and member head or sink is decreased. The dynamic cluster algorithm combines with prediction of trajectory can prolong network life under accuracy of target tracking...