SHARP: A Study on Safe Heterogeneous Agent Reinforcement Learning Paradigm for 5G-TSN Traffic Scheduling

With the rapid development of the Industrial Internet of Things (IIoT), 5G-TSN (5G-Time Sensitive Network) has emerged as a promising solution in IIoT to address stringent requirements for low latency and high reliability in industrial networks. However, it introduces a critical challenge in the synchronization of coordinated decision-making between heterogeneous devices for optimizing end-to-end traffic scheduling. To tackle this issue, we propose a heterogeneous multiagent service scheduling optimization architecture for 5G-TSN networks and develop a strategy based on Safe Heterogeneous Agent Reinforcement Learning Paradigm (SHARP). Our approach uniquely combines coordinated decision-making among heterogeneous network elements, adaptive scheduling balancing critical and non-critical traffic, and a safety mechanism ensuring reliability for time-sensitive flows. Extensive simulations demonstrate that SHARP significantly outperforms existing algorithms, achieving improved industrial network throughput and higher acceptance rates for time-triggered flows, thus advancing the efficiency and reliability of industrial communication systems.