Cognitive Power Control Under Correlated Fading and Primary-Link CSI

We consider the cognitive power control problem of maximizing the secondary throughput under an outage probability constraint on a constant-power constant-rate primary link. We assume a temporally correlated primary channel with two types of feedback: perfect delayed channel state information (CSI) and one-bit automatic repeat request (ARQ). We use channel correlation to enhance the primary and secondary throughput via exploiting the CSI feedback to predict the future primary channel gain. We provide a numerical solution for the power control optimization problem under delayed CSI. In order to make the solution tractable under ARQ-CSI, we re-formulate the cognitive power control problem as the maximization of the instantaneous weighted sum of primary and secondary throughput. We propose a greedy ARQ-CSI algorithm that is shown to achieve an average throughput comparable to that attained under the delayed-CSI algorithm, which we solve optimally.