On the Impact of Node Failures and Unreliable Communications in Dense Sensor Networks

We consider the problem of decentralized detection in failure-prone tree networks with bounded height. Specifically, we study and contrast the impact on the detection performance of either node failures (modeled by a Galton-Watson branching process) or unreliable communications (modeled by binary symmetric channels). In both cases, we focus on ldquodenserdquo networks, in which we let the degree of every node (other than the leaves) become large, and we characterize the asymptotically optimal detection performance. We develop simple strategies that nearly achieve the optimal performance, and compare the performance of the two types of networks.