Research On WSN Tracking Algorithm Based On Dynamic Clustering

Wireless Sensor Networks(WSNs) is a network of application, it was initiallyintended for military, deployed in battlefield to detect the information of enemy. With thecontinuous development of technology and the progress of the society, more andmore civil public application introduce the technology of WSNs.WSNs has been appliedin many aspects of life, such as traffic management, health care, smart home, publicsafety, danger detection, environmental protection and so on. Furthermore, technology ofmicro-electronics and nanometer is developing, sensors are more smaller andmultifunction, so that we can infer that the applications of WSNs will continue to expandin the future, and WSNs has broad prospects for development as well.Target tracking is one of important applications of WSNs，originally used to trackthe enemy forces, then traffic tracking, track animal for research, medical patient trackingand so on are the important areas of application. Along with the continuous improvementof the algorithm, the requirement is higher and higher, people pay more attention to theprediction of target motion trajectory. The prediction of target motion trajectory is toinform sensors in prediction area for preparation before the target accesses, so that mostof the sensors can save energy instead of keeping detecting. Now it is generally acceptedthat application of clustering technology is better for management and maintenance of thenetwork. After combining clustering technology and tracking, target tracking algorithmbased on clustering is published. However, the tracking prediction puts forward morerequirements on clustering. It is hot topic of research in which matter can make it morereasonable for clustering management and tracking on the premise of prediction accuracy.First the structure and characteristics of WSNs is introduced in this paper, moreovertarget tracking research background and significance, then propose the currently majorchallenges for target tracking. We introduce the principle and evaluation index of targettracking, analyze the existing classic target tracking algorithm. Then we introduce severalkinds of available clustering algorithms, they can be divided into two classes: one isclustering with nothing else, the other is clustering combined with tracking, does not apply to other applications. After researching the tracking algorithm and clusteringalgorithm, we propose a target tracking algorithm based on dynamic clustering. Thealgorithm do not need central control, or predicting the target trajectory by massivecalculation in sensors. The process of clustering and tracking is executing in sensors byself, without extra aid and maintenance. Algorithm considers the speed and direction oftarget movement, the sensors outside cluster are into the ready state in advance. Thenthey decide whether take part in the competition of cluster by judge the angle of targetmovement, and decide which layer of sensors is more reasonable by judge the speed oftarget movement. Finally algorithm combines distance and residual energy, calculateweight value associated competition wait time, choose the appropriate node to becomecluster head. Guarantee the tracking accuracy and durability.We validate the algorithm by computer simulation in various aspects of performance.We come to a conclusion that algorithm can use energy properly, balance the sensors’load, decreases the number of communication data packet between sensors, so that it willprolong the network survival time, be more applicable to civil large-scale network.