.png){kind=icon}
.png){kind=icon}
Return
Predictive QoE-Driven Radio Resource Management via Network Digital Twin in 5G and Beyond Networks
.png){kind=icon}
OA
AI
E
A
G
G
DOI：10.1109/tnsm.2026.3716878.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
The evolution toward Beyond 5G networks introduces stringent requirements for intelligent Radio Resource Management (RRM) capable of jointly optimizing Quality of Experience (QoE) and resource utilization under highly dynamic conditions. This paper proposes a predictive QoE-driven RRM framework built upon an AI-enabled Network Digital Twin (NDT), which operates as a high-fidelity replica of the physical network to support proactive and efficient system-level resource allocation. The proposed approach integrates a Deep Learning (DL)-based module for forecasting future objective QoE metrics, namely Mean Opinion Score (MOS) values, with a Deep Reinforcement Learning (DRL) agent for dynamic Physical Resource Block (PRB) allocation. By incorporating predicted QoE levels over a finite horizon into the DRL agent’s state representation, the framework enables foresighted and policy-aware resource management while reducing synchronization overhead between the NDT and the physical infrastructure. Extensive simulations under heterogeneous traffic loads and QoE policies demonstrate that the proposed approach maintains high median QoE levels while adaptively regulating resource utilization, avoiding the systematic saturation observed with baseline static schedulers. The results further highlight stable learning behavior across DRL variants and confirm real-time feasibility with limited computational overhead.
Keywords:
Quality of experience (QoE)
radio resource management (RRM)
network digital twin (NDT)
artificial intelligence (AI)
5G and beyond networks
Journal
IF:
5.4
Papers:
509
Citations:
9.2K
Organization
Cited Papers