Efficient Multi-Mechanism Coordination for Time-Sensitive Networking: Interrelation-based Iterative Optimization Method

The rapid expansion of Industrial Internet of Things applications necessitates on-demand communication for largescale heterogeneous flows. According to their differential demands, Time-Sensitive Networking offers a promising scheme combining multiple appropriate mechanisms like Cyclic Queueing and Forwarding (CQF) and Credit-based Shaper (CBS). Existing research focuses on the unidirectional impact analysis between multi-mechanisms but neglects their interrelation to optimize resource allocation, significantly limiting the network scheduling. For this reason, this paper establishes a MultiMechanism Coordinative Scheduling (MMCS) architecture that integrates CQF&CBS to efficiently schedule large-scale heterogeneous flows on-demand. Through the three phases of flow scheduling, mechanism co-analysis, and scheduling arbitration, MMCS iteratively generates, assesses, and adjusts the scheduling solutions by providing feedback from the latter two phases to the first phase. To precisely assess the solution generated by flow scheduling phase, a tighter worst delay is derived as the mechanism co-analysis metric by modeling the impact of CQF on CBS. Guided by the assessment result, a scheduling arbitration strategy is devised to balance the allocation of scheduling resources by reversely adjusting CQF&CBS parameters, enabling the next iteration of flow scheduling phase. The experiments show that MMCS dramatically improves the efficiency and schedulability of heterogeneous flow scheduling, extending scheduling capacity from existing 1000 to over 10000 flows.