Skip to Main Content
In this work, we consider the sensor localization problem from a novel perspective by treating it as a functional dual of target tracking. In traditional tracking problems, static location-aware sensors track and predict the position/speed of a moving target. As a dual, we utilize a moving location-assistant (LA) (with global positioning system (GPS) or pre-defined moving path) to help location-unaware sensors to accurately discover their positions. We call our proposed system Landscape. In Landscape, an LA (an aircraft, for example) periodically broadcasts its current location while it moves around or through a sensor field. Each sensor collects the location beacons, measures the distance between itself and the LA based on received signal strength (RSS), and individually calculates their locations via an unscented Kalman filter (UKF) based algorithm. Our contributions are at least twofold. (1) Landscape is a distributed scheme, it does not rely on measured distances among neighbors (as used by most current proposals), which makes it robust to topology and density; Landscape involves zero sensor-to-sensor communication overhead, and is highly scalable to network size. (2) By introducing UKF in sensor localization problem, we reap multiple benefits: our UKF-based algorithm nicely exploits the constraints increasingly added by the beacons; it elegantly solves the nonlinear problem with low computation cost and complexity; and most importantly, it efficiently reduces the effects of measurement errors, making Landscape robust to ranging errors. Extensive simulations and evaluations against the state-of-the-art systems show that Landscape is a high-performance sensor positioning scheme for outdoor sensor networks.