Broadly interpreted, sensor management denotes the theory and application
of dynamic resource allocation in a diverse system of sensors and sensing
modalities. Over the past few years the problem of sensor management for
active sensing has received an increasing amount of attention from researchers
in areas such as signal processing, automatic control, statistics, and machine
learning. Active sensing is recognized as an enabling technology for the next
generation of agile, multi-modal, and multi-waveform sensor platforms to efficiently perform tasks such as target detection, tracking, and identification. In a managed active sensing system, the sequence of sensor actions, such as pointing angle, modality, or waveform, are selected adaptively based on information extracted from past measurements. When the adaptive selection rule is care fully designed such an on-line active approach to sensing can very significantly improve overall performance as compared to fixed off-line approaches. However,due to the classic curse of dimensionality, design and implementation of optimal active sensing strategies has been and remains very challenging.
Recently, several U.S. research funding agencies have supported efforts in
areas related to sensor management. There has also been interest in operations
research approaches to sensor management, e.g., multi-armed bandits. This
has led to research activity in academia, government laboratories, and industry
that has borne some fruit in specific technology areas. For example, several
promising new methods to approximate optimal multistage sensor management
strategies for target tracking have been developed and an understanding
of design challenges and performance tradeoffs is beginning to emerge.