Power-constrained sensor selection and routing for cooperative detection in cognitive radios

Given a spectrum-sensing network, a set of active nodes jointly aggregate sensed data at a preset frequency-band and simultaneously route this information to an arbitrarily chosen querying node through a power-constrained multi-hop path. Locally, each sensor node is assumed to be an energy-based detector. This work focuses on deriving algorithms that jointly optimize sensor selection and cooperative detection from which a power-efficient route to the querying node can be established, and then, a tree routing structure spanning the chosen nodes is constructed under a power budget constraint. Sensor information is sequentially aggregated along this optimized routing structure up to the querying node. Each parent node combines the information coming from each of its child nodes using either log-likelihood ratios or optimal linear weights. This is done with the goal of maximizing, at the querying node, the overall probability of detection (P_D) for a given probability of false alarm (P_FA) and a given total power budget spent by the sensor network for routing the sensor information. We propose two algorithms: (1) greedy and (2) select-aggregate-forward that provide a trade-off between the detection quality and the power consumption. We provide experimental results that show the outperformance of our algorithms against traditionally proposed routing structures such as the shortest-path-tree.