LAUSR

Device-free localization (DFL) enables several new applications in various sectors including smart cities, intelligent transportation, and public safety. DFL relies on a network of sensor radars that transmit, receive, and process reflected signals propagating in a monitored environment. The accuracy of DFL degrades in cluttered environments, due to the presence of undesired objects that reflect the signal. Indeed, the multiple reflections of the signal overlap at the receiver and make the inference of targets’ positions challenging. This article presents a theoretical foundation of DFL in cluttered environments by deriving the fundamental limits on DFL accuracy. In particular, we propose a system model that takes into account multiple reflections, nonline-of-sight conditions, and the presence of multiple targets. Building on such a model, we derive the Cramér-Rao bound on the inference accuracy of targets’ positions by applying equivalent Fisher information analysis. The proposed bound provides guidelines for the design and analysis of DFL systems operating in cluttered environments. Then, the article presents a case study compliant with the 5G New Radio numerology and channel modeling. Results show how the minimum achievable error is affected by multiple reflections and multiple targets and to which extent the employment of a signal with larger bandwidth and a network with a higher number of receivers can lower the achievable error toward submeter accuracy.