Joint Node Association and Beamforming for Millimeter Wave Networks with Multiple Reconfigurable Intelligent Surfaces

Reconfigurable intelligent surface (RIS) is a promising technology to increase the spatial degrees-of-freedom in sparse millimeter wave (mmWave) channels. In RIS-assisted mmWave networks with multiple nodes, however, optimization of node-association and beamforming (BF) may require a large channel training overhead. This paper proposes a joint node-association and cooperative BF scheme to maximize the weighted sum rate with a low overhead based on a two-step design. In the first step, only slowly varying channel information are used to determine the node-association, analog active BF, and passive BF. In the second step, the baseband BF and power allocation are designed only using the effective channel information with reduced dimension. It is shown that the proposed scheme outperforms benchmark schemes and is scalable to the numbers of nodes.