Geometry-based algorithms for device-free localization with wireless sensor networks

In this paper, geometry-based algorithms for device-free localization (DFL) of a single target are proposed. The algorithms exploit the change in attenuation caused by the device-free target on radio links passing through the deployment area. By solving for the intersections of some of the attenuated links and taking the centroid of the intersection points, an estimate of the target's location is obtained. To increase the accuracy, only a subset of the top attenuated links are considered in the location estimation. Furthermore, a moving average scheme is utilized to reduce errors due to the variation in received-signal strength (RSS) caused by environmental factors. This removes the need for extensive calibration of baseline RSS as used in other DFL schemes, and is a more robust approach compared to using the RSS measurement from the previous sampling period as the baseline RSS. Experimental results obtained using the proposed algorithms with a Kalman filter are promising, with 2.09 ft tracking RMSE.