O-M³: Real-Time Multi-Cell MIMO Scheduling in 5G O-RAN

Open radio access network (O-RAN) enables cooperative signal processing among multiple cells at a centralized O-RAN distributed unit (O-DU). It is a key technology for cellular networks to increase spectrum efficiency. To achieve cooperative signal processing across multiple cells, a new scheduler is needed. Specifically, the scheduler must jointly determine RB allocation, MCS assignment, and beamforming matrices for all users from all the cells that are involved in multi-cell processing. In addition, the scheduler must obtain its scheduling solution within each TTI (i.e., at most 1 ms) to be useful for the frame structure defined by 5G NR. In this paper, we present O- $\mathbf M^{3}$ —a real-time scheduler for multi-cell MIMO networks under the O-RAN architecture. O- $\mathbf M^{3}$ can meet the stringent timing requirement with joint optimization of beamforming matrices, RB allocation, and MCS assignment among multiple cells. O- $\mathbf M^{3}$ is developed through a novel multi-pipeline design that exploits parallelism. Under this design, one pipeline performs a sequence of operations for cell-edge users to explore joint transmission, and in parallel, the other pipeline is performed for cell-center users to explore MU-MIMO transmission. We implement O- $\mathbf M^{3}$ on a commercial off-the-shelf (COTS) GPU. Experimental results show that O- $\mathbf M^{3}$ is capable of offering a scheduling solution within 500 $\mu \text{s}$ for an O-RAN system with 7 O-RAN radio units (O-RUs), 100 users, 100 RBs, and $2\times 8$ MIMO. O- $\mathbf M^{3}$ can also meet the 1 ms requirement for $2\times 12$ MIMO systems. Meanwhile, O- $\mathbf M^{3}$ can provide ~40% throughput gain on average through joint transmission across multiple cells.