Indoor Moving Target Localization Based On Wireless Sensor Networks

Location service has significant value in lots of areas, such as military, industry, civil, disaster relief and so on. However, the widely used Global Positioning System (GPS) can hardly be used in indoor environment. Wireless sensor network (WSN) provides a new solution to indoor target localization, but the performance of WSN is susceptible to environmental noises. It’s necessary to design more efficient localization algorithm to improve the positioning accuracy. This thesis mainly studies on the localization and tracking algorithms of moving target in indoor WSN environment, which have significant academic value and practical significance.Most of the current indoor localization algorithms are implemented based on the prior knowledge of environment noise, which is difficult to obtain in practical application. Aiming at the situation that the environment noises and measurement noises is unknown, a particle swarm optimization (PSO) based minimum residual (minRE) localization algorithm is proposed. The localization task is transformed into an objective function which represents the minimum residual, and can be solved with the PSO algorithm. Experiment results show that the proposed PSO-minRE can increase the accuracy for at least 20%.Considering the line-of-sight (LOS) and non-line-of-sight (NLOS) condition coexist and randomly transition in complex indoor target tracking, an interacting multiple model (IMM) distance filter is proposed to smooth and compensate measurement errors under LOS/NLOS. Then, the extended kalman filter (EKF) is employed to localize the target with the filtered distance. Experimental results illustrate that the IMM filter can efficiently mitigate the NLOS ranging errors and achieve high localization accuracy with mean location error of 1.2m.For three dimensional indoor environment which is more practical, a data-fitting (DF) based least square (LS) tracking algorithm is proposed. With the proposed algorithm, the historical information of the LS localization results are effectively used to optimize the current LS localization result. Simulation results show that the proposed algorithm’s positioning accuracy can reach lm without knowing the prior knowledge of the environment noise. Therefore, the proposed algorithm is of great practical significance.Finally, the research work of this paper is summarized and the further research directions are given.