Research On Optimazition Methods In Clustered Sensor Networks

As a new information acquisition and processing method,wireless sensor networks(WSN) are widely used in fields such as military investigation, environmental monitoring and health care.WSN is data-based that intermediate nodes can fuse the received data,while nodes in traditional wireless networks are only responsible for relaying packets.Because wireless sensors are usually deployed in hostile environments,the battery energy can not be replenished,thus protocols for WSN must be energy efficient.Traditional protocols are not fit for WSN.In WSN, data collected by neighboring nodes are often similar,cluster based routing protocols can effectively simplify the query for WSN,lower communication cost and save energy by letting cluster heads fuse data sensed by its member nodes before relaying. Cluster based routing protocols meet the requirements of data-based routing,and are more robust and scalable.This paper focuses on the study of energy efficient optimization methods at network layer and MAC layer for cluster based WSN.The research contents are:Firstly,this paper discusses the optimization of join-cluster-model for cluster based routing protocols at network layer.Clustering algorithms with uniform radius guarantee the scalability of cluster heads.However,neighboring clusters are often overlapped,i.e.,member nodes are usually covered by more than one cluster heads. One principle for clustering is to balance the load among all clusters.In this paper,a cluster size estimation model,CSEM,is proposed to balance to load among clusters. The major drawback of this model is the heavy extra energy consumption.Thus another model:a load balanced join cluster metric,namely LBJCM,which requires low communication cost for cluster heads is proposed.Simulation results show that CSEM and LBJCM both outperform other common join cluster models.Secondly,this paper discusses the study of optimizing the TDMA schedule algorithms at MAC layer for cluster based WSN.Contention based MAC protocols cause much collision,thus waste nodes' energy.TDMA based schedule can greatly save nodes' energy.One typical TDMA schedule method for cluster-based WSN is that each cluster head allocates different time slots for its member nodes,and collects data in a round-robin way(e.g.the LEACH protocol).To avoid the inter cluster interference,current solutions include assigning a unique DSSS code for every cluster,and assigning different channels for neighboring clusters.However, these solutions requires extra hardware support,thus are not suitable for the heavily dispersed WSN scenarios.An interference free cluster based TDMA schedule protocol,IFCT,solves the problem.IFCT protocol employs a two-leveled schedule to avoid transmission interference.The first schedule assigns different frame number to neighboring clusters so that neighboring clusters could collect their members' data during different frames,avoiding inter-cluster interference.The second schedule allocates different slots to cluster members and they sending their data to their respectively cluster heads during different slots to avoid intra-cluster interference. The major draw back of IFCT is that the frame assignment algorithm FA of IFCT heavily wastes channel utilization.Focusing on this problem,this paper proposes two new schedule algorithms:ETS and ETS2.Both of them have a low time and message cost complexity and greatly improve the average channel usage radio under different cluster radius.Thirdly,this paper proposes a data fusion oriented routing protocol based on LEACH,DF-LEACH.Because cluster heads communicate with base station directly in one hop,those nodes which are far from base station die quickly.Thus,LEACH is improved by proposing a model to estimate the location of the outburst event.Each cluster head estimates the location of the sensed event using this model,then finds out the next hop node.Simulation results show that DF-LEACH prolongs the network lifetime comparing with LEACH.Finally,a fault-tolerant scheduling algorithm upon the MAC layer is implemented in a real wireless meter system project.Wireless meter system has a strict restriction for the meter packets loss ratio.A distributed scheduling algorithm is used to schedule collectors to report their meter data.Experiment results show that all packets are successfully collected.