Slice Resource Management With MADDPG-Based Traffic Classification in 5G/B5G Networks

To address the limitations of 5G/B5G network slicing approaches in meeting diverse and high-reliability service demands, this paper proposes the Slice Resource Management based on Traffic Classification and Multi-agent Deep Deterministic Policy Gradient (SRM-TCM) framework. The SRM-TCM framework leverages Multi-agent Deep Deterministic Policy Gradient (MADDPG) for centralized training and dynamic resource scheduling, enabling precise traffic classification and adaptive slice management. This approach enhances resource utilization and ensures high Quality of Service (QoS), outperforming traditional methods that lack traffic-specific resource optimization.