Intelligent User Association and Scheduling in Open RAN: A Hierarchical Optimization Framework

In the ever-evolving landscape of <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">NextG</i> wireless networks, Open radio access network (RAN) emerges as a transformative paradigm, revolutionizing network architectures and fostering innovation through its open, intelligent and disaggregated approach. By integrating RAN intelligent controllers (RICs), we can seamlessly implement machine learning (ML) algorithms to cater to diverse vertical applications and deployment environments without the need for intricate planning. However, this architecture suffers from two critical challenges: frequent handovers and load balancing amid varying traffic demands of different services in dynamic environments. To address these issues, this study proposes a joint intelligent user association, congestion control, and resource scheduling (IUCR) scheme. Aligning with the 7.2x functional split (FS) option recommended by the O-RAN Alliance, we present a hierarchical optimization framework incorporating heuristic methods, successive convex approximation (SCA), and a distributed deep reinforcement learning (DRL) approach across different Open RAN components, such as RICs and RAN layers. The simulation results convincingly demonstrate the superior performance of the proposed scheme compared to centralized approaches, validating its effectiveness.