On flexible association and placement in disaggregated RAN designs

In Open RAN architectures, the classic gNB radio protocol stack is disaggregated into virtualized components: the Centralized Unit (CU), the Distributed Unit (DU), and the Radio Unit (RU). Each unit is deployed throughout a cloud-enabled RAN infrastructure to meet users’ Quality of Service (QoS) requirements. In this framework, we propose Open RAN unit placement methods that maximize User Equipment (UE) admission while ensuring their QoS needs. We focus on two primary tasks: (i) establishing UE-RU associations and (ii) placing CUs and DUs across the network’s cloud hosts. We formulate the joint association-placement UE-DU-CU optimization problem as an Integer Linear Programming (ILP) model and propose two resolution approaches besides the optimal one: (i) an algorithm that decomposes and sequentially solves the ILP model and (ii) a Recurrent Neural Network (RNN) heuristic that emulates the joint optimization model. We assess the optimal model’s performance across varying network resource availability. Compared to baseline models, simulations demonstrate that our approaches ensure higher admissibility levels while minimizing deployment costs and increasing fairness. The RNN heuristic presents a small optimality gap, with up to 9% fewer admissions while reducing the execution time by up to 99.98%, making it suitable for real-time implementation.