Date of Award


Degree Type



Electrical and Computer Engineering

First Advisor

Zhu, Ye

Subject Headings

Blind source separation, Sensor networks, Wireless communication systems, Automatic tracking, Signal processing, Tracking, Wireless sensors, BSS algorithms


This thesis describes an approach to track multiple targets using wireless sensor networks. In most of previously proposed approaches, tracking algorithms have access to the signal from individual target for tracking by assuming (a) there is only one target in a field, (b) signals from different targets can be differentiated, or (c) interference caused by signals from other targets is negligible because of attenuation. We propose a general tracking approach based on blind source separation, a statistical signal processing technique widely used to recover individual signals from mixtures of signals. By applying blind source separation algorithms to mixture signals collected from sensors, signals from individual targets can be recovered. By correlating individual signals recovered from different sensors, the proposed approach can estimate paths taken by multiple targets. Our approach fully utilizes both temporal information and spatial information available for tracking. We evaluate the proposed approach through extensive experiments. Experiment results show that the proposed approach can track multiple objects both accurately and precisely. We also propose cluster topologies to improve tracking performance in low-density sensor networks. Parameter selection guidelines for the proposed topologies are given in this Thesis. We evaluate proposed cluster topologies with extensive experiments. Our empirical experiments also show that BSS-based tracking algorithm can achieve comparable tracking performance in comparison with algorithms assuming access to individual signals