Node Selection Based Target Localization/Tracking Algorithm And Its Application In Wireless Sensor Networks

With the fast technology development in wireless communication, micro-electro-mechanical systems (MEMS), as well as sensing techniques, wireless sensor networks (WSNs) have found a wide range of applications. WSNs usually comprise of small and relatively inexpensive sensor nodes, which are capable of collecting, processing, storing and transferring the information of the monitored environment. WSNs can be employed in applications like environmental supervising, battlefield surveillance and status monitoring and maintenance applications. Among all different kind of these applications, target localization and tracking is one of the vital research focuses for WSNs. How to improve the accuracy and energy efficiency, however, challenges the moving target tracking.Firstly, the target localization and tracking model will be introduced in this thesis, and some popular target localization algorithms are reviewed for comparison. After an analysis of the existing localization algorithms, the vertical localization algorithm was introduced, which determines the location of the target by the vertical line crossing between neighboring sensor nodes which observe the presence of the target. The critical factors which dominate the achieved localization accuracy are revealed to motivate the need for node selection in vertical line location algorithm. And the node selection criteria together with the node selection procedure are proposed in this thesis to enable the node selection based vertical line localization algorithm with improved accuracy. Moreover, the influence of inaccurate fading factors and the background noise on the achieved localization performance are highlighted. The simulation results are presented to validate that, when comparing with other localization techniques which do not need the distance measurement between sensor node and target, the vertical line localization may achieve reasonable localization performance, thus offering a promising alternative for the target localization application in WSNs.Secondly, the moving target tracking problem and the state-of-the-art techniques in moving target tracking are reviewed. More specifically, the thesis focuses on the closest point of approach (CPA) based target tracking technique, wherein the moving target position, velocity, moving direction and continuous trajectory in the effective monitoring region could be traced by using the CPA information from neighboring sensor nodes. Compared with other algorithms, CPA algorithm is able to trace a reasonable continuous trajectory close to the actual target moving trajectory. In addition, CPA algorithm only requires that the CPA times of multiple sensor nodes along the moving target trajectory are stored and exchanged among sensors, so the message exchange overhead is low. Then the thesis also discusses the influence of sensor node selection on the target tracking performance to highlight the importance of node selection in CPA algorithm. A node selection criterion was proposed to enhance the localization and tracking performance of CPA algorithm. On the other hand, a sliding window method is proposed to alleviate the noise contamination on the CPA time measurement, thus improving both the accuracy and the robustness of CPA algorithm in the noisy environment. Simulation platforms over the OPNET are presented to enable the numerical simulations of ECPA, improved ECPA with node selection and the vertical line localization. It is unveiled through simulation results that, the improved ECPA algorithm with node selection outperforms the ECPA algorithm in terms of the localization accuracy. Meanwhile, sliding window averaging pre-processing is effective in suppress the noise deterioration.Finally, based on the energy consumption analysis of CPA algorithm, a dynamic cluster structure was proposed to improve the energy efficiency in moving target tracking and localization in WSNs. And the simulation results are presented to validate that, the dynamic cluster mechanism could be utilized to reduce the communication message number which corresponds to the reduced energy consumption, thus prolonging the lifetime of the WSNs with the premise of the successful target tracking.