Abstract. The wireless local area networks represent an alternative to custom sensors and dedicated surveillance systems for target indoor detection. The availability of the channel state information has opened the exploitation of the spatial and frequency diversity given by the orthogonal frequency division multiplexing. Such a fine-grained information can be used to solve the detection problem as an inverse scattering problem. The goal of the detection is to reconstruct the properties of the investigation domain, namely to estimate if the domain is empty or occupied by targets, starting from the measurement of the electromagnetic perturbation of the wireless channel. An innovative inversion strategy exploiting both the frequency and the spatial diversity of the channel state information is proposed. The targetdependent features are identified combining the Kruskal-Wallis test and the principal component analysis. The experimental validation points out the detection performance of the proposed method when applied to an existing wireless link of a WiFi architecture deployed in a real indoor scenario. False detection rates lower than 2 [%] have been obtained. Recently, the channel state information (CSI) used by WLANs in the framework of orthogonal frequency division multiplexing (OFDM) has been exploited to enable location-based services. The CSI is a fine-grained feature of the channel frequency response (CFR) providing both magnitude and phase information at multiple frequencies [11]. This indicator is highly sensitive to the environmental changes and its information content can be exploited to characterize the target signature on the electromagnetic field. The multiple-input multiple output (MIMO) architecture has been introduced by the current WLAN standards to further improve the quality of the data transmission. The intrinsic spatial diversity of a MIMO system has been also exploited to enhance the performance of the wireless localization problem.
IntroductionMany techniques exploiting the prominent features of the CSI exist in the state of the art, but the indoor scenario and the system setups are often dedicated to the localization itself. On the contrary, the proposed technique aims to introduce the detection capability on top of an existing wireless architecture already used for communication.