Joint Routing and Injection Time Planning Algorithm Based on CQF Shaper

Time-Sensitive Networking (TSN) provides deterministic communication guarantees for real-time applications through various traffic scheduling mechanisms. The Cyclic Queuing and Forwarding (CQF) mechanism, introduced by the IEEE 802.1Qch standard and known for ensuring end-to-end bounded latency, has become a focal point of research. Current studies on CQF highlight issues such as the separation of routing and traffic injection slot scheduling, leading to low schedulability of flows and resource utilization. In this paper, we propose a traffic scheduling method based on joint routing and injection time planning. By jointly considering routing and injection time planning, the overall scheduling scheme is optimized. In the routing phase, a routing algorithm assisted by reinforcement learning calculates a set of candidate paths for the flow and prioritizes them based on path information to select routes with richer slot resources. During the scheduling phase, the method assigns scheduling priorities based on traffic characteristics and schedules according to the set of paths from the routing algorithm, thus expanding the solution search space and aiming to maximize the number of schedulable flows. Experimental results demonstrate that this method significantly increases the number of schedulable flows while ensuring the end-to-end average latency of time-sensitive flows.