Joint Unicast and Multicast Transmission in RSMA-Aided Clustered Cell-Free Systems: Clustering and Power Allocation

In this article, we design a joint unicast and multicast transmission system based on rate-splitting multiple access (RSMA) and cell-free (CF) architectures. To enhance multicast efficiency and system scalability, we propose a large-scale fading coefficients-based clustering scheme that partitions users and access points (APs) into disjoint clusters, achieving dual benefits: significant precoding complexity reduction and enhanced multicast transmission efficiency. Unlike conventional methods, our approach enforces a critical constraint ensuring the number of cluster transmit antennas exceeds user antennas, thereby guaranteeing sufficient independent spatial streams. Subsequent to cluster formation, we implement fully distributed power allocation, substantially reducing system overhead compared to centralized alternatives. We derive closed-form expressions for the average sum rate (ASR) and ergodic sum rate (ESR) lower bounds under distributed operation, and develop a graph convolutional network (GCN) power allocation algorithm. Simulation results validate the superiority of our approach.