On Scheduling With Resource Allocation in Multi-Provider Cell-Free Networks

Abstract

We study resource allocation in multi-provider open radio access network (O-RAN) cell-free networks and analytically prove that allocating bandwidth proportionally to each cenrral unit (CU)’s time-averaged per-user queue length is the optimal solution to a coarse-timescale Lyapunov drift minimization problem, under the condition that a MaxWeight-like scheduler equalizes queue lengths within each CU. To implement this framework, we propose a two-tier hierarchical architecture in which the Non-Real-Time RAN intelligent controller (RIC) executes the proportional bandwidth rule over multiple slots, while individual CUs independently run a low-complexity per-slot user-association algorithm based on the MaxWeight criterion. Simulation results show that the proposed proportional rule can effectively reduce queue backlogs by dynamically redistributing bandwidth under various settings.