Multiple target tracking by a distributed UWB sensor network based on the PHD filter

Abstract

In this paper we describe a method for detection, localization and tracking of multiple non-cooperative moving targets by multiple static ultra-wideband (UWB) radar sensors. The targets are detected using electromagnetic waves backscattered by the target towards the receivers. The sensors consist of two receiving and one transmitting antenna connected to a single UWB module, making the sensors capable of autonomous target localization. Gaussian mixture implementation of the probability hypothesis density (PHD) filter is used for estimating the ranges of the detected targets and fusion of the location estimates provided by each sensor, resulting in refined target tracks. The proposed method is verified in a realistic scenario with three moving persons. The results show that UWB sensors can be used as complimentary technology for multiple moving target tracking.