Research On Target Tracking In Wireless Sensor Networks With Mobile Elements

Wireless sensor networks consist of numerous sensors, which are widely used for collecting environmental information. Target tracking is an important application for Wireless Sensor Networks(WSNs). Its main objective is to monitor the targets in the area of interest by sensors in the network. For traditional fixed sensor networks, there are several problems in target tracking such as high energy consumption, existing coverage holes and so on. With the development of electronic technologies and manufacturing techniques, more and more mobile elements, i.e., mobile sensors, have been used in practical applications. With their mobility, mobile sensors are very suitable for tracking targets: on the one hand, mobile sensors can follow the target and then monitor it effectively; on the other hand, mobile sensors are not fixed in their initial positions and can move in the area of interest, which can effectively eliminate coverage holes in the network. By surveying other literatures, the author found that there are also several problems in existing detection-centric target tracking methods with mobile elements. First, most of these methods focused on mobile sensor networks which consisted of mobile sensors only. They hardly considered the cooperation between mobile nodes and fixed nodes. Second, most of them used the unit disk model for describing nodes' sensing ability. This kind of model cannot reflect the transmitting situation of signals. Third, the issue of multi-target tracking was rarely considered in these studies.To solve these problems, the author do the research on target tracking in heterogeneous(hybrid) sensor networks with mobile elements. The heterogeneous sensor networks consist of fixed nodes and mobile nodes. In this paper, fixed nodes and mobile nodes can cooperate with each other in order to monitor targets. And a realistic signal attenuation model is used to describe sensors' measurements. Moreover, different algorithms are designed for single target tracking and multiple targets tracking, respectively. The main research work in this paper is listed as follows:1) Target detection method in homogeneous networks. Target detection is a preliminary for target tracking. In this paper, there are no thresholds for signal intensity. Sensors' measurement values are converted into conditional probabilities. Based on this, a probabilistic method for decision making is proposed.2) Single target tracking method in hybrid sensor networks with mobile elements. A few of mobile sensors are introduced in the fixed sensor network to form a heterogeneous sensor network. In the proposed method, mobile nodes are scheduled to get close to the target and monitor it in collaboration with fixed nodes.3) Multiple targets tracking method in hybrid sensor networks with mobile elements. In order to schedule the movement of mobile nodes reasonably and try not to miss any target, a virtual action force model is designed. Based on it, the algorithm can ensure the tracking quality of all targets as much as possible.4) Single target tracking method in heterogeneous sensor networks with mobile camera sensors. Based on virtual monitoring region, a cascading tracking method is proposed to schedule mobile camera sensors for monitoring the target.5) Multiple targets tracking method in heterogeneous sensor networks with mobile camera sensors. A metric value is designed for allocating mobile camera sensor among all targets. Based on it, a multi-target tracking method is proposed.Based on above research work, this paper proposed several different target tracking methods for the target tracking problem in heterogeneous sensor networks. These methods can be used for scenarios of single target and multiple targets, respectively. In order to validate those proposed methods, the author conducted extensive simulations. The results show that the use of mobile sensor can improve tracking quality effectively. In conclusion, the research work not only fills in blanks of existing methods, but also provides practical value for reference.