Enhancing the Performance of Propagation Model-Based Positioning Algorithms

Abstract

Object localization in wireless networks through Received Signal Strength (RSS) measurements requires a precise estimation of the signal attenuation model in order to produce meaningful results. The popular lognormal channel model, widely adopted to describe the signal strength attenuation as a function of the distance between nodes, turns out to be too simplistic when applied to a real scenario. In this paper, we analyze two possible improvements to this model: on one hand, we build a different channel model for each reference node in the network, with the aim of tackling the anisotropy of the environment. On the other hand, we explicitly append to the lognormal model a term to account for walls attenuation. A thorough experimental testbed demonstrates the potentials of the two approaches, with the second one being especially useful to counteract the effect of the limited sensitivity of practical wireless receivers.