Hierarchical beamforming for downlink fog RAN

A scenario consisting of several cells is considered where each cell contains a multiple-antenna enhanced remote radio head (eRRH) and several single-antenna mobile stations (MSs). All eRRHs can access the cloud with limited capacity fronthaul links. The beamforming problem aiming at maximizing the sum rate with per eRRH power constraint is considered. In this paper, a hierarchical one-shot top-down approach that decomposes the beamforming problem between the cloud and eRRHs is proposed. The basic idea of our approach is that each eRRH leaves out part of its transmit signal space for minimizing the inter-cell interference to other cells while exploiting the rest of the signal space termed transmission subspace for data transmission. This idea is elaborated in a hierarchical order. First, the cloud, using only the knowledge of link qualities among eRRHs and MSs averaged over all antennas, finds the size of the transmission subspace at every eRRH. Using this information, each eRRH finds the basis of its transmission subspace and correspondingly optimizes its beamforming vectors employing only local channel knowledge. The results show that feeding back only link qualities to the cloud, a significant performance gain can be achieved compared to uncoordinated eRRH based approaches. Furthermore, our proposed approach achieves sum rates close to the global maximum at all signal to noise ratios.