Device-Free Indoor Localization Based on Ambient FM Radio Signals

Introduction

A large body of research has been dedicated to localization of mobile devices. However, in certain situations the user does not (or cannot) carry any device (for example, smart environments, assisted daily living). Device-free localization (DFL) addresses the challenge of locating the user without any wearable devices. The existing DFL techniques have a number of limitations which prevent their wide adoption. In particular, computer vision methods require installation of multiple cameras, which need particular light conditions, infringe users' privacy and cause psychological discomfort. Another DFL technology, pressure-sensitive floors, while preserving user privacy, have prohibitive cost of deployment. Radio-based DFL methods, in turn, are based on effects of radio wave interactions with human body (such as diffraction, reflection, scattering), which ultimately result in measurable changes of signal properties (Scholz et al., 2011). These methods typically require low-cost hardware, allow hidden installation, and cannot violate users' privacy as vision-based systems do. State-of-the-art radio-based DFL systems require deployment of multiple wireless devices which actively transmit, receive and analyse radio signals (Wilson & Patwari, 2011; Scholz et al., 2011; Kosba et al., 2012). Due to radio spectrum regulations, these devices are typically restricted to use only one or few frequency channels in a narrow license-free band. This paper investigates feasibility of DFL with a single indoor receiver employing ambient FM radio stations. In contrast to related work, the proposed system is completely passive (since it uses ambient transmitters) and monitors multiple radio channels simultaneously. Multi-frequency scanning is the key feature of the system; in this paper, several FM radio stations are monitored simultaneously with a software-defined radio (rtl, 2013). Experimental results demonstrate that channel diversity can considerably improve localization performance and thus enable sub-room level localization with ambient stations. The following sections review related work, introduce the proposed approach, and present experimental results for coarse-grained and fine-grained localization. The paper concludes with a summary of findings.