RSS-based clustering of mobile terminals for localization in wireless networks

Received signal strength (RSS) metric has been attracting a lot of interest in localization in wireless systems since it is available by default. However, this metric is affected by a shadowing phenomenon that highly degrades the localization accuracy. In this paper, we show that the RSS can be exploited for performing a clustering operation and finding the groups of geographically nearby mobile terminals. We propose a localization solution that does not directly estimate the position from the RSS, but instead performs clustering, and then computes the position of a mobile terminal as a function of the available positions of other mobile terminals belonging to its cluster. We solve the clustering problem by means of a non-parametric technique. We also treat the problems of removing the clutter that corresponds to spatially isolated mobile terminals and filling the missingness in the RSS vectors. Simulations illustrate the performance of the proposed solution in a macro cellular deployment scenario.