Map-aware RSS localization models and algorithms based on experimental data

The knowledge of the propagation environment can greatly improve the performance of indoor wireless localization systems, since it can be exploited to cancel out, at least to some extent, the effects of physical obstructions (e.g., walls) degrading the radio signals employed for localization. In indoor localization systems, the propagation environment can be described by maps (e.g., floor plans), which represent a valuable source of knowledge. In this paper a new map-aware statistical model for the measurements acquired in wireless localization systems based on received signal strength is proposed. Experimental results are processed to extract the parameters of the proposed model, which is then exploited to devise map-aware localization algorithms; these are capable of mitigating the non-line-of-sight bias introduced in the measurements by the obstructions modelled by the map. Our numerical results evidence that map-aware modelling can substantially improve localization accuracy in indoor scenarios with respect to map-unaware modelling.