Optimal Stochastic Coordinated Beamforming with Compressive CSI Acquisition for Cloud-RAN

Cloud radio access network (Cloud-RAN) is a revolutionary RAN architecture. It provides a cost-effective way to improve both the network capacity and energy efficiency by shifting the baseband signal processing to a single baseband unit (BBU) pool, which enables centralized signal processing. However, in order to exploit the performance gains of full cooperation, full channel state information (CSI) is often required, which will incur excessive signaling overhead and degrade the network performance. To resolve the CSI challenge for Cloud-RAN, we propose a novel CSI acquisition method, called compressive CSI acquisition. This new method can effectively reduce the CSI signaling overhead by obtaining instantaneous coefficients of only a subset of all the channel links. As a result, the BBU pool will obtain mixed CSI consisting of instantaneous values of some links and statistical CSI for the others. We then propose a stochastic coordinated beamforming (SCB) framework to deal with the uncertainty in the available mixed CSI. The SCB problem turns out to be a joint chance constrained program (JCCP) and is known to be highly intractable. In contrast to all the previous algorithms for JCCP that can only find feasible but sub-optimal solutions, we propose a novel stochastic DC (difference-of-convex) programming algorithm with optimality guarantee. To reduce the computational complexity, we also propose two low-complexity algorithms using the scenario approach and the Bernstein approximation method for larger-sized networks. Simulation results will show that the proposed compressive CSI acquisition method can reduce the CSI overhead significantly, and the proposed SCB algorithms provide performances close to the full CSI case.