blending sensor scheduling strategy with particle filter to track a smart target混合传感器调度策略与粒子滤波跟踪一个聪明的目标.pdfVIP

Wireless Sensor Network, 2009, 1, 300-305 doi:10.4236/wsn.2009.14037 Published Online December 2009 (/journal/wsn). Blending Sensor Scheduling Strategy with Particle Filter to Track a Smart Target 1 1 2 3 Bin LIU , Chunlin JI , Yangyang ZHANG , Chengpeng HAO 1 Department of Statistical Science, Duke University, Durham, U. S. A 2Adastral Park Research Campus, University College London, London, UK 3Institute of Acoustics, Chinese Academy of Sciences, Beijing, China Email : {bin.liu2, chunlin.ji }@, y.zhang@adastral.ucl.ac.uk, haochengp@ Received April 17, 2009; revised July 20, 2009; accepted July 21, 2009 Abstract We discuss blending sensor scheduling strategies with particle filtering (PF) methods to deal with the prob- lem of tracking a ‘smart’ target, that is, a target being able to be aware it is being tracked and act in a manner that makes the future track more difficult. We concern here how to accurately track the target with a care on concealing the observer to a possible extent. We propose a PF method, which is tailored to mix a sensor scheduling technique, called covariance control, within its framework. A Rao-blackwellised unscented Kal- man filter (UKF) is used to produce proposal distributions for the PF method, making it more robust and computationally efficient. We show that the proposed method can balance the tracking filter performance with the observer’s concealment. Keywords: Particle Filter, Sensor Scheduling, Smart Target, Tracking 1. Introduction sensor to