Fens G035,22 Temmuz 2011,11:00-12:30
Adaptive sensor management strategies for wireless sensor networks (WSNs) determine the optimal way to manage system resources and task a group of sensors to collect measurements for statistical inference. In this introductory level talk, I’m going to present several adaptive sensor management strategies for event detection, source localization and target tracking using quantized sensor data. I’ll first present results on sensor threshold design using multi-objective optimization. I’ll then talk about static source localization and compare two sensor selection metrics based on mutual information and Fisher information in terms of mean squared error (MSE) and computational complexity.
Finally, I’m going to introduce our recent work on dynamic bit allocation for target tracking. Since the optimal solution maximizing the determinant of the Fisher information matrix (FIM) subject to the total bandwidth constraint requires a combinatorial search, we seek computationally efficient suboptimal methods. In order to maximize the determinant of the FIM, we formulate an approximate dynamic programming algorithm and compare its performance with other suboptimal methods in terms of MSE and computational complexity as well.
BIO: Engin Maşazade got his B.S. degree from Electronics and Communications Engineering department from Istanbul Technical University in 2003. He then obtained his M.S. and PhD. degree in 2006 and 2010 respectively from Sabanci University, Electronics Engineering Program. Since 2008, he is also affiliated with the Sensor Fusion Group at Syracuse University, department of Electrical Engineering and Computer Science where he is now a post-doctoral researcher. His research interests include distributed detection, localization and tracking for wireless sensor networks.