Wireless Sensors in Distributed Detection Applications

Detection problems provide a productive starting point for the study of more general statistical inference problems in sensor networks. In this article, the classical framework for decentralized detection is reviewed and argued that, while this framework provides a useful basis for developing a theory for detection in sensor networks, it has serious limitations. The classical framework does not adequately take into account important features of sensor technology and of the communication link between the sensors and the fusion center. An alternative framework for detection in sensor networks that has emerged over the last few years is discussed. Several design and optimization strategies may be gleaned from this new framework