An Indoor Localization Algorithm Based on Modified Joint Probabilistic Data Association for Wireless Sensor Network

Due to the localization accuracy of global positioning system (GPS) cannot meet the requirements in indoor environment, the wireless sensor network (WSN) techniques are efficient methods to cope with this problem. The WSN-based indoor localization techniques have become effective methods to solve the problem of indoor localization. Since the nonline-of-sight (NLOS) effect could severely induce the localization accuracy, the primary challenge in indoor localization is the handling of NLOS errors. Due to multipath effect, near-far effect, obstacle occlusion, etc., the NLOS errors become very sophisticated. Aiming at this problem, a modified joint probabilistic data association localization (MJPDA) algorithm is proposed in this article. First, MJPDA obtains a series of preprocessing virtual points by grouping the measurements. Then, the measurements are divided into two categories, that is, line-of-sight (LOS) and NLOS, by virtual points density. In the case of LOS, extended Kalman filter (EKF) is used for processing. For the NLOS case, a series of particles are first generated around the prediction point, and then modified JPDA is used to data association of the virtual points and the particles. Simulations results illustrate that MJPDA is superior to MPDA algorithm and the traditional EKF algorithm in localization accuracy and robustness. Finally, we perform the real experiment to verify the performance of MJPDA. The experimental results demonstrate that MJPDA has prominent performance in mitigating large NLOS errors.

Automated summary

Indoor localization faces challenges caused by NLOS effects. The MJPDA algorithm improves localization accuracy and robustness by grouping measurement results and performing data association.