Research On Device-Free Localization Based On RF Signal And Channel State Information

Device-free localization (DFL), which does not require any device attached to target, is attractive, such as intrusion detection and elderly monitoring. Compared with camera or infrared based solutions, radio-frequency (RF) based device-free localization approaches can work at day and night, and also can penetrate nonmetallic walls.A lot of RF-based device-free localization methods have been proposed, however, those methods have limitations in real-world application. First, to achieve high localization ac-curacy, most recent DFL methods rely on collecting a large number of Received Signal Strength (RSS) changes distorted by target and incurred high energy consumption. Second, the past proposes assume a fixed distribution of the RSS changes even they are distorted by different types of targets. It inevitably causes the localization to fail if the targets for modeling and testing belong to different categories. Third, most of existing device-free lo-calization schemes use the received signal strength to localize a target, they suffer from low accuracy in rich multipath environment. Fourth, existing device-free localization systems, however, either require labor-intensive offline training or dedicated hardware, which results high cost in reality. Thus, it is important to find a more general device-free localization method for real-life application.Based on the signal distortion caused by target, this paper analysed the impact of the target on the signal amplitude change, the signal phase change and the channel state information change. To achieve low energy consumption, target adaptive, high localization accuracy and low human cost, This paper proposed four kinds of localization systems for four different application scenarios. The main contribution of this paper is given as follow:(i) To deal with the high energy consumption problem caused by large amount of data col-lection, this paper introduces a compressive sensing (CS) based energy-efficient and high-precision multi-target DFL method. Compared with the existing methods, the proposed method demands fewer transceivers. The motivation is the sparse nature of multi-target locations, then this paper applies CS to guarantee high accuracy with less RSS change mea-surements. This paper also introduces an adaptive orthogonal matching pursuit algorithm to recover the location vector with unknown target number.(ii) To cope with the high human cost problem caused by target diversity, this paper presents a transferring compressive sensing based DFL approach, which employs a rigor-ously designed transferring function to transfer the distorted RSS changes across different categories of targets into a latent feature space, where the distributions of the distorted RSS changes from different categories of targets are unified. Thus, the same transferred sensing matrix can be shared by different categories of targets, leading to a reduction of the human efforts.(iii) To solve the low localization accuracy caused by multipath effect, this paper introduces D-Watch, a device-free localization system built on top of low cost commodity-off-the-shelf (COTS) hardware. Unlike previous works which consider multipaths detrimental, D-Watch leverages the "bad" multipaths to provide a decimeter-level localization accuracy without any offline training. D-Watch harnesses the angle-of-arrival (AoA) information from the Radio-frequency identification (RFID) tags'backscatter signals. The key intuition is that whenever a target blocks a signal's propagation path, the signal power experiences a drop and this power change can be accurately captured by the proposed novel P-MUSIC algorithm.(iv) To deal with the high deployment cost problem caused by labor-intensive offline train-ing, this paper presents a model-based device-free localization system LiFS, implemented on cheap COTS WiFi devices. Unlike previous COTS device-based work, LiFS is able to localize a target accurately without offline training. The challenge is that the RSS or even the fine-grained channel state information (CSI) can not be easily modelled due to rich mul-tipath indoors. We observe that even in a rich multipath environment, not all subcarriers are affected equally by multipath reflections. By identifying the subcarriers not affected by multipath, CSIs on the "clean" subcarriers can be utilized for accurate localization.