Exploiting opportunistic interactions for localization in heterogeneous wireless systems

Abstract

This paper addresses the problem of localization in mobile networks. Our goal is to make localization possible even for off-the-shelf communication devices like smartphones or sensor nodes, not equipped with GPS or operating in areas where the GPS does not work well. In our approach, a user estimates its position by exploiting opportunistic exchanges with other devices (peer devices). The localization information provided by peers include their own position estimation and can be used even though it may be highly inaccurate. In a heterogeneous setup, peers can provide different ranging technologies (e.g. RSSI and UWB). We investigate the performance of two existing localization algorithms based on Weighted Centroid (WC) and Linear Matrix Inequalities (LMI) under different conditions of accuracy and heterogeneity. The study is performed by means of simulations that, however, make use of realistic ranging models, derived from an extensive set of RSSI and UWB measurements. The simulation results show that in most cases LMI provides a better user position estimation than WC, with an error of 1 to 4 meters for 10 opportunistic interactions, and that heterogeneity of peer positioning accuracy has a limited but positive impact on the localization performance.