enDRTS: Deep Reinforcement Learning Based Deterministic Scheduling for Chain Flows in TSN

With the rapid development of artificial intelligence (AI) technology, learning-based time-sensitive networking (TSN) technology can be used as a promising network technology to facilitate automated network configuration in the industrial Internet of Things (IIoT). However, the stricter multiple requirements of the IIoT have posed significant challenges, that is, deterministic and bounded latency for jointly chain transmission of multiple service flows. In this paper, we propose an enhanced scheduling algorithm, namely enDRTS, based on deep reinforcement learning (DRL) for chain transmission of chained service flows to jointly solve the time slot scheduling problem in TSN. By analyzing the feature of flows and scheduling constraints, the DRL algorithm model can adjust the queue bandwidth allocation and transmission slot strategy of switch ports in time to make enDRTS more salable and efficient. Simulation experiments are conducted to evaluate the performances of enDRTS, and the results show that enDRTS can schedule more flows and improve time slot utilization compared with benchmark methods.