IEEE/SP 13th workshop on Statistical Signal Processing July, 17-20, 2005 - Bordeaux - France

All papers by session

Information Theoretic Approaches to Adaptive Sensing and Sensor Management
Plenary Lecture 1

Abstract

Sensor management combines resource allocation and signal processing in a powerful way that allows successive sensor actions, e.g. where to point a sensor or what sensor modality to use, to be optimally determined based on information extracted from previous measurements. The sensor management problem arises in many applications including sensor networks, wireless communications, security, and medical diagnostics. Driving sensor actions according to information theoretic formulations of the problem is a popular method that is gaining ground over other task-driven approaches. In this talk we will discuss information theoretic approaches to sensor management as compared and contrasted to task driven methods, i.e., methods that select some physical performance criteriion and explicitly manage the sensor based on that criterion. Illustrations and examples will be drawn from sensor networks and sensor swarms, multi-modality target tracking, and other sequential design of experiments problems.