Title: Two-tier target tracking framework in distributed sensor networks
Authors: Lufeng Shi; Jindong Tan
Addresses: Department of Electrical and Computer Engineering, Michigan Technological University, Houghton, MI 49931, USA ' Department of Mechanical, Aerospace and Biomedical Engineering, University of Tennessee, Knoxville, TN 37996, USA
Abstract: A distributive two-tier tracking framework is developed to track mobile targets in sensor networks. The algorithm utilises the efficiency of binary sensor reading to ensure fast isolation of the target, and then it uses a small cluster of sensors around the target to compute accurate target location. The framework has two tiers, in tier-one, binary sensor reading and Viterbi algorithm are used, it is a fast and distributed way of roughly locate the target. In tier-two, a detailed localisation using maximum likelihood estimation is done within a small cluster of sensors to compute the accurate location of the target. By dividing the framework into two tiers, scaling the size of the network only affects the computation cost in tier one, where binary data are used. The computation cost in tier 2 remains unchanged. The framework is first implemented in a chained-form network with sensors deployed in a regular pattern, it is then extended to other regularly deployed network in full 2D space. The two-tier tracking framework is implemented using ultrasonic sensor network in a real-world hallway-room environment in our lab.
Keywords: distributed sensor networks; target tracking; two-tier framework; mobile targets; sensor clusters; target location; binary sensor reading; Viterbi algorithm; localisation; maximum likelihood estimation; ultrasonic sensors.
DOI: 10.1504/IJSNET.2014.065784
International Journal of Sensor Networks, 2014 Vol.16 No.1, pp.32 - 40
Received: 19 Jan 2013
Accepted: 19 Mar 2014
Published online: 19 Nov 2014 *