The Distributed Location Tracking System Based On Wireless Sensor Networks

As a new attractive IT technology, wireless sensor networks are very promising in many military and civil applications. Target tracking is among the most typical and challenging applications of the wireless sensor networks (WSNs). There is a contra-diction between the extreme complexity of target tracking and the low capacity of each sensor node. Distributed is an essential characteristic of WSNs, which is also a powerful measure to solve this contradiction. Through the collaborative work among all neighboring nodes, the lack of powerful capacity for a single node is overcame in order to satisfy the application requirements. The fundamental theories, key technologies and system implementations are studied in this dissertation.First of all, due to the fact that the network coverage is an important precondition for the target tracking, the method to measure the network coverage quality is studied given an arbitrary node distribution, especially to measure the "path exposure" for target tracking. The Grid Scan algorithm is proposed for the area coverage issue, which is then extended to the barrier coverage issue. Moreover, the greedy node redeployment algorithms are proposed to improve the quality of coverage for the area coverage and barrier coverage issues, respectively.In this dissertation, the target tracking problem is investigated according to the logical sequence: ranging,localization and tracking. Based on the analysis of the error source for current acoustic ranging methods, two acoustic ranging algorithm are proposed, SDE-FD-TDoA and SDE-ToA~2 , based on sound source delay elimination and frequency detection. The former one has a better accuracy and the latter doesn't need time synchronization. Although having the advantages such as better accuracy and simple localization algorithm, ranging-based localization methods are suffering from the disadvantages such as scalability, hardware cost and directionality. So, the range-free localization methods based on RF signals are studied. A weighted centroid localization method, W-Centroid, is proposed, which is based on RSS and connectivity metrics. W-Centroid has some advantages such as simple and easy to use, low hardware cost, acceptable localization accuracy.Then, for the outdoor environment, a distributed RF-based target tracking system NemoTrack is proposed. With the motion of the target, the nodes near the target dynam- ically form an autonomous tracking group by leader electing, which is used to localized the target, while the nodes far away from the target enter the sleep state that can improve the network life effectively. A tiny database is running on the leader node, which is used to store and manage the data reported by the member nodes. The time efficiency and space efficiency of this tiny database are both high, which is especially suitable for solving the "reporting storm" problem caused by reporting to the leader node by members at the same time. Based on these reported data, the leader node aggregates these data using the weighted centroid localization algorithm, which can decrease the communication traffic significantly. A reliable multihop routing protocol is proposed based on the minimal cost tree and local parents switching, in order to transfer the localization results reliably from the leader node to the sink node. Then, a Kalman filter is used to filter out the noise and get a smoother trajectory for the target. NemoTrack has some advantages such as good scalability, no need for additional hardware and low power design.At last, with regard to the indoor environment, a distributed RF-based indoor local-ization system iNemo is proposed, which has hierarchical localization accuracy, room-level separation and cube-level positioning. Through the interaction among the neigh-boring beacons, the radio signal strength (RSS) maps are built automatically without any complicated manual signal collection phase. Based on the fluctuation of RSS between the beacon and the target, a PCI (Positioning Confidence Indicator) metric is proposed to represent the confidence of the localization results.Among above four sections, three sections are validated using experiments on the popular Mica2 motes except section 1, which uses experiments based on Matlab simulation tools.