In cognitive radio networks, it is challenging for secondary users (SUs) to estimate and control their interference at the receivers of primary users (PUs), due to incomplete or erroneous channel information between SUs and PUs. Thus, SUs need to estimate the worst-case aggregate interference at PU receivers to ensure guaranteed protection for PUs from excessive interference. As it is rare that all SU-PU channels experience the worst-case conditions simultaneously, we propose a practical model (namely, the worst-case selective robust model) for SUs to estimate their aggregate interference power. This model employs an adjustable parameter to control the number of SU-PU channels that are in the worst-case conditions.
For an individual SU-PU channel, the estimation of worst-case channel gain is subject to a distribution uncertainty. Given this robust model, we study SUs’ power control problem in a non-cooperative game where each SU selfishly maximizes its own throughput performance subject to coupled interference constraints at PU receivers. We study the existence and uniqueness of Nash equilibrium and propose an iterative algorithm for SUs to achieve the equilibrium in a distributed manner. Numerical results show that our algorithm provides guaranteed protection for PUs and fair throughput performance for SUs, provided with uncertain SU-PU channel information.