Previous Abstract Return to Session C4 Next Abstract

Session C4: Alternative/Terrestrial-based Opportunistic PNT

Sensors Fusion using RSS based Particle Filter in Precise Indoor Localization
Ruojun Li, Liyuan Xu, Julang Ying,Kaveh Pahlavan, Worcester Polytechnic Institute
Location: Windjammer
Alternate Number 3

Recently, it is a trend that various sensors in a smartphone are fused in the popular Wi-Fi localization. Received signal strength (RSS) is a common candidate for indoor localization, which is easy to collect and process, but it is significantly affected by shadow fading, making it fail to estimate the location accurately. On the other hand, sensors like accelerometer, gyroscope, barometer are widely used in pedestrian dead reckoning (PDR) tracking, which provides a possibility of fusing RF and other sensors’ signal for precise localization and tracking.
In this paper, we propose a particle filter algorithm to optimize the PDR result, followed by a Weighted Path Loss Algorithm(WPL). In the experiment, we collected data from the 3rd floor on Atwater Kent Laboratory at Worcester Polytechnic Institute. After optimized by our method, a significant improvement in localization precision is achieved. Comparing the performance of this algorithm with four popular Wi-Fi localization algorithms (Kernel, Weighted centroid, KNN, and LMS), the results show that this algorithm outperforms the others. Further exploration is also made to examine the effect of number, deployment and dropping strategies of particles and numerical results are also illustrated in the comparison of localization error and computational complexity.



Previous Abstract Return to Session C4 Next Abstract