Optimal bit and power allocation for rate-constrained decentralized detection and estimation

In decentralized wireless sensor networks, the decision taken at the fusion center depends on the quality of the signals received from the sensing nodes. The typical sources of errors are observation noise, quantization errors due to source encoding, and communication noise, i.e. errors in the transmission from the sensing nodes to the sink node. In this work we propose a method for allocating quantization bits and transmission powers in each node in order to satisfy alternative optimization criteria in either detection or estimation. In both cases, to limit transmission errors, we enforce the additional constraint that the number of coding bits on each sensor is less than the channel capacity over the channel that connects each sensor to the sink node.