A New Approach to Energy Efficient Signal Detection

Signal detection is one important application for which dedicated sensor networks have been proposed for both civilian and military applications. Joint signal processing and communication design of such networks has been of great interest recently. Several authors have made a strong case for a mode of operation called censoring in order to reduce the number of sensor transmissions such that energy usage will be reduced to allow longer network operation. There are some aspects of wireless sensor networks that have not yet been exploited by the proposed censoring schemes. For example, sufficiently close nodes can hear each others' transmissions. Such considerations lead to new classes of energy efficient signal detection procedures which we consider here. We demonstrate that these new approaches require fewer average sensor transmissions and that the savings can be significant in cases of interest. We demonstrate our new approach is optimum for minimizing a weighted sum of probability of error and the average number of sensor transmissions for a distributed mode of operation where each sensor node can operate autonomously.