Key points are not available for this paper at this time.
The Proximity Radio Access Network (P-RAN) has emerged as a promising architecture for extending coverage and enhancing throughput through user-assisted relaying. However, user mobility and dynamic channel conditions cause frequent link disruptions and interference, posing challenges to reliable transmission and efficient resource allocation. This paper proposes a learning-based two-stage optimization framework for joint relay selection and wireless resource allocation in P-RAN systems. The objective is to maximize system sum rate while improving transmission reliability under dynamic network conditions. We first derive the outage probability and formulate a stochastic mixed-integer nonlinear programming (MINLP) problem that jointly considers relay selection, subcarrier assignment and power allocation. To address computational intractability, we develop a two-stage scheduling scheme in which machine learning is employed for intelligent relay selection, followed by semidefinite relaxation and Lagrange multiplier methods for power and subcarrier optimization. Simulation results show that the proposed approach significantly improves system sum rate and reduces outage probability compared with conventional heuristic algorithms, achieving near-optimal performance with lower computational complexity. These findings demonstrate that the proposed framework provides an effective and scalable solution for real-time resource management in dynamic P-RAN environments and offers valuable insights for future 6G network design.
Niu et al. (Wed,) studied this question.