Robust optimization of cognitive radio networks powered by energy harvesting

We consider a cognitive radio network, where primary users (PUs) share their spectrum with energy harvesting (EH) enabled secondary users (SUs), conditioned on a limited SUs´ interference at PU receivers. Due to the lack of information exchange between SUs and PUs, the SU-PU interference channels are subject to uncertainty in channel estimation. Besides channel uncertainty, SUs´ EH profile is also subject to spatial and temporal variations, which enforce an energy causality constraint on SUs´ transmit power control and affect SUs´ interference at PU receivers. Considering both the channel and EH uncertainties, we propose a robust design for SUs´ power control to maximize SUs´ throughput performance. Our robust design targets at the worst-case interference constraint to provide a robust protection for PUs, while guarantees a transmission probability to reflect SUs´ minimum QoS requirements. To make the non-convex throughput maximization problem tractable, we develop a convex approximation for each robust constraint and successfully design a successive approximation approach that converges to the global optimum of the throughput objective. Simulations show that SUs will change transmission strategies according to PUs´ sensitivity to interference, and we also exploit the impact of SUs´ EH profile (e.g., mean, variance, and correlation) on SUs´ power control.