Detecting the Direction of Motion in a Binary Sensor Network

Abstract

We examine the problem of detecting the direction of motion in a binary sensor network; in such a network each sensor’s value is supplied reliably in a single bit of information: whether the moving object is approaching towards or moving away from the sensor. We demonstrate that the geometric properties of the network itself can be exploited for the detection of movement direction, from a single instance of sensor reading only. Moreover the estimation is performed in a distributed processing fashion, with only a minimal data collection at situation-dependent leading sensors and features a low computational burden on each sensor. In addition, different detection instances drain the resources of different groups of sensors, of a small size compared to the size of the whole network. Our experiments demonstrate high accuracy that increases with sensor density and/or sensing range, while the responsiveness of the detection model is practically instantaneous.