Ubiquitous Radio Sensing: Localization of Non-Cooperative Users

Abstract

Many applications in the emerging smart environments, such as smart cities and intelligent transportation, require an accurate situational awareness, i.e., information about presence and location of all users in the environment. For example, situational awareness enables location-aware home control, assistive health, or vehicular safety applications. A reliable situational awareness is based on robust communications and sensing techniques, as well as large-scale deployment with respect to the application environment. This demand for both communications and sensing capabilities has recently led to significant research interests in merging the two technologies allowing for an efficient use of spectral resources and a beneficial mutual assistance. An integration of sensing functionality is envisioned as a native capability of next generation wireless networks. The deployment of these perceptive wireless networks ultimately offers the prospect of ubiquitous radio sensing. Ubiquitous radio sensing allows, in particular, the localization of non-cooperative users. In this context, non-cooperative users are defined as users who are neither equipped with a communications device nor with a dedicated localization device. State-of-the-art algorithms on localizing non-cooperative users typically obtain the required location information exclusively from directly scattered or reflected signals. Further signal components resulting from multipath propagation, for example due to reflections from the surrounding environment, are considered as interfering signals, i.e., clutter, and are suppressed by the receiver. In this thesis, we propose to exploit the full potential of signal propagation, including multipath, for the localization of non-cooperative users. Therefore, we first provide detailed insights into time-based radio sensing. We derive, develop, and implement the fundamental performance bounds, clutter mitigation techniques, as well as parameter estimation and tracking algorithms to localize non-cooperative users. Moreover, we provide a novel radio sensing approach, referred to as multipath-enhanced device-free localization (MDFL), which complements time-based radio sensing. Instead of mitigating signal components, MDFL makes use of the spatial information contained in the corresponding propagation paths to localize non-cooperative users. For localization, MDFL exploits user-induced variations in the received power of the individual signal components. These signal components include besides the line-of-sight also multipath components. For realizing this novel sensing approach, we develop corresponding measurement models relating the user-induced variations in the received power of the individual signal components to the user’s location. Based on these measurement models, we derive the fundamental performance bounds for MDFL. Finally, we present a Bayesian formulation of the sensing problem and implement corresponding non-parametric filter solutions for MDFL, which are successfully applied to measurement data. To the best of the author’s knowledge, this is the first time it is shown that variations in the power of multipath components can be used for the localization of non-cooperative users. Due to its low computational complexity, the proposed MDFL approach represents a suitable technique for the emerging perceptive wireless networks and can help to enable ubiquitous radio sensing.