Adaptive statistical sampling methods for decentralized estimation and detection of localized phenomena

The design and deployment of sensor networks (SNET) for decision making pose fundamental challenges due to energy constraints and uncertain environments. In this paper we focus on one such problem where minimization of communication costs due to information exchange is required subject to end to end information quality constraints. Specifically, we develop solutions for detection of distributed events, sources, or abnormalities that are localized, i.e., only a small number of sensors in the vicinity of the phenomena are in the field of observation. This problem complements the standard decentralized detection problem, where noisy information about a global event is measured by the entire network. The global phenomena by itself can be one of several different discrete possibilities and researchers have investigated several architectures within this context. Our objective in this paper is to characterize the fundamental tradeoffs between global performance (false alarms and miss rate) and communication cost. We develop a framework to minimize the communication cost subject to worst-case misclassification constraints by making use of the false discovery rate (FDR) concept along with an optimal local measure transformation at each sensor node. The preliminary results show that the FDR concept applied in a sensor network context leads to significant reduction in the communication cost of the system.