Fairness Analysis in IRS assisted C-RAN with Imperfect CSI

Abstract

Intelligent reflecting surfaces (IRS) are being considered as a potential technology for the future genereations of wireless networks. IRS is a low cost and energy efficient technology to boost the performance of existing wireless systems by providing some control over the propagation channel. In this paper, we focus on fairness in cloud radio access network (C-RAN) and investigate the impact of integrating IRS into the system. In particular, as attaining the full channel state information (CSI) is difficult in IRS systems, we evaluate the performance of IRS assisted C-RAN with imperfect CSI. To ensure the fairness amongst all users, we choose maximizing the minimum expected user rate as the optimization problem. The problem is shown to be stochastic and non-convex which is computationally prohibitive. We propose an algorithm that jointly optimizes the beamformers and the IRS phase shifts. A statistical coordinated descent (SCD) optimization is used to maximize the minimum ergodic user rate. To deal with stochasticity of the optimization problem, we utilize the sample average approximation (SAA) along with weighted minimum mean square error (WMMSE) methods. Finally, the numerical results are presented. They show that particularly at low signal to noise ratio (SNR) regimes, deploying IRS can help increase the maximized minimum rate significantly.