Abstract
In terms of the location-aware applications for wireless sensor networks, it is crucial to improve the measurement reliability of the received physical signal as well as the target positioning and tracking performance in a real environment. In the localization process, reflection, scattering, and other physical phenomena have a negative impact on the measurement reliability and the estimation accuracy at local sensor nodes. In this paper, we describe using a particle filter to increase the received signal strength (RSS) measurement reliability of each local sensor node and then use the least mean squares (LMS) algorithm to estimate the path loss exponent of the distance-dependent path loss model in our laboratory environment. Using the Industrial Technology Research Institute (ITRI) sensor platform, the experimental results show that the proposed LMS-based method can reduce the average noise variance by approximately 1 dB. Then the filtered RSS samples and the estimated path loss exponent are applied in the proposed weighted sign algorithm (WSA), which are implemented in the ITRI sensor platform for target positioning and tracking in our laboratory surroundings. The location estimation of a target is formulated as a weighted least squares (WLS) problem, and then is solved based on the WSA in an iterative and decentralized manner. Experimental results demonstrate that the proposed WSA-based positioning and tracking scheme achieves a maximum error distance of approximately 1.7 meters, average error distances of approximately 0.96 meters (positioning case) and 1.38 meters (tracking case), which are not far from those obtained from computer simulations (0.67 meters and 0.91 meters, respectively). ©2010 IEEE.