A WMMSE-Based Contiguous Resource Scheduling Algorithm for 5G-NR Uplink

In 5G NR, contiguous resource allocation, in which each user must be allocated with a contiguous subset of resource blocks (RBs), is expected to be followed for practical uplink scheduler to reduce the signaling overhead for scheduling. Due to the spatial-domain coupling caused by the multi-user MIMO (MU-MIMO) transmission and frequency-domain coupling caused by the contiguous resource allocation constraint, finding the optimal contiguous resource allocation in 5G-NR uplink requires exhaustive search that has exponential complexity w.r.t. the number of users and RBs. This letter presents a weighted-minimum-mean-square-error (WMMSE)-based contiguous resource scheduling algorithm, which alternatively optimizes the receive matrix, mean squared error (MSE) weight matrix and resource allocation until convergence. The resource allocation subproblem can be further decomposed into independent resource allocation subproblems for each user, whose optimal solution can be found within polynomial complexity. A low complexity resource allocation method is proposed to further reduce the complexity to satisfy the real-time scheduling requirement in 5G-NR. Experimental results show that the proposed algorithms can achieve much better scheduling performance than existing baselines with comparable complexity.