A Prototype Ultrasonic Sensor Network for Tracking of Moving Targets

This paper describes the design and implementation details of our recently developed prototype target tracking system using ultrasonic sensor network. It reports our findings of inter-sensor interference, and suggests a dime-division based sensor scheduling strategy to avoid it. Extended Kalman filter (EKF) algorithm is proposed as the tracking algorithm with the characteristic of variable time difference between successive time steps and variable measurement equation due to different sensors used at different time steps. A maximum likelihood data association algorithm is presented for multi-target tracking. In addition, a series of important issues including the hardware setup, software architecture, synchronization mechanism, database design, statistics for tracking and alerting management, visualization etc. are also introduced. Comparisons between different tracking algorithms are given, and the experimental results for tracking of multiple targets are demonstrated