Research On Localization Algorithm And Tracking Algorithm Based On Wireless Sensor Networks

Wireless sensor networks (WSN), which integrates the technologies of sensor, micro-electro-mechanism system (MEMS), wireless communication and distributed computing, is a novel technology for acquiring and processing information. It connects the physical world to the logical information world, and will have a profound influence on the life style of mankind in 21st century after internet technology. As a new technology, WSN promises many new application areas, such as military application, counter-terrorism application, anti-riot application, emergency application, health application, environmental application, intelligent home application and other commercial applications.Sensor node localization and target tracking are two vital problems in the research area and application area of WSN. Worth the whistle, the localization problem is the foundation and premise of the target tracking. However, the unrenewable and constrained power of sensor node, and the infertile circumstance where WSN is deployed bring great challenges in localization application and tracking application. In order to meet the requirements of WSN, the sensor node localization problem and target tracking problem were analyzed from the views of precision, robustness and energy consumption based on the existing localization mechanisms and tracking mechanisms. The major contribution of this dissertation is specifically stated as follows:(1) A distributed localization algorithm (DMDS-DC) for WSN was proposed based on multidimensional scaling and distance calibration. In DMDS-DC, the local positioning regions (LPR) were established by an adaptive search algorithm, which not only kept that the relative coordinates of sensors can be transformed between neighbor LPRs, but also reduced the number of redundant common sensors in network. In each LPR, the shortest path distances between sensors was corrected through the SPDCG algorithm and SPDCWV algorithm. Using the SMACOF (scaling by majorizing a complicated function) algorithm and the classical MDS algorithm, the relative positions of sensors were computed and optimized. When the global relative positions or absolute positions of sensors were required, these can be obtained by merging relative maps of all LPRs based on the information of anchors and common sensors. Simulation results showed that the DMDS-DC algorithm is an energy-efficient algorithm with small localization error and has strong robustness to range errors in different sensor deployments.(2) A prediction-based distributed target tracking algorithm (PBD-ER), which coupled with a error recovery mechanism, was proposed. In PBD-ER, the future location of target was predicted by the PFSS algorithm which can capture more moving patterns at the same condition. Then, complying with the principle of increasing the sleeping time of sensors, the target was tracked by the dynamic wakeup cluster (DWC). If the location prediction error was large or the location can not be predicted by PFSS algorithm, the target was monitored by only a sensor which is nearest to the target. At the reporting time, all sensors in DWC would wakeup and sensed the distances to the target. Based on these distances, the location of target was calculated by head node of DWC through DMDS-DC algorithm. When the loss of target occurred during the process of tracking, a three levels recovery mechanism, which activated the range of sensors gradually, was applied to find the target. Simulation results showed that the PBD-ER algorithm can achieve better performance on tracking precision, prediction precision, energy efficiency and robustness to range error.(3) A energy efficient and cluster-based event boundary tracking algorithm (EBTAC) was proposed. In EBTAC algorithm, the whole event boundary was split according to the static clusters in network. In each cluster which can sense the event, a closed curve was formed by the cluster boundary and event boundary. Then the nodes of event boundary (NEB) were found based on the closed curve and the EBNSA algorithm. In EBNSA algorithm, an agent was employed to search all the NEBs. During the process of searching, the agent didn't exchanged information with the interior sensors of the event and only exchanged information with the sensors around the event boundary to obtain the NEBs. In order to increase the coverage rate of NEBs to the event region, the external event point (EP) was introduced and the calculation method of the EP was also discussed in detail. Simulation results showed that the EBTAC algorithm can track the event boundary with high precision, low energy consumption, and low memory requirement, and the EPs can also be used as a substitute for the NEBs to achieve a better coverage rate to the event region.