Adaptive signaling schemes for detection in wireless sensor networks

Abstract

A binary decentralized detection problem in which sensor nodes provide partial information about their environment to a fusion center is studied. The nodes have access to conditionally independent observations and transmit a summary of their own data over wireless channels. Upon reception of the information, the fusion center attempts to accurately reconstruct the state of nature. The communication link between each node and the fusion center is modeled as a fading channel corrupted by additive noise. Channel state information is available at the fusion center and at the sensor nodes. Large deviation theory is used to show that having identical sensor nodes is asymptotically optimal. Algorithms in which each sensor node selects a signaling/coding schemes based on the quality of its channel are studied.