Deterministic Cognition: Cross-Domain Flow Scheduling for Time-Sensitive Networks
Guoyu Peng, Shuo Wang, Zongquan Li, Tao Huang, Chaowei Yuan
Abstract
With Industry Internet being more advanced and sophisticated, different closed network domains tend to collaborate on the same task and provide strict Quality of Service (QoS) guarantees. However, current Time-Sensitive Networking (TSN) and Deterministic Networking (DetNet) technologies make it hard to meet the needs of multi-domain collaboration with bounded low service latency. To deal with this dilemma, a time-sensitive flow scheduling scheme across multiple network domains is necessary. In this paper, we propose a deterministic cross-domain flow scheduling (DCDFS) solution to overcome the barriers of inter-domain microbursts not considered by TSN and DetNet. First, a novel hierarchical architecture is designed to support cross-domain coordination, deterministic cycle mapping, and flexible queue stitching functions for DCDFS. Second, we build the system model and formulate the cross-domain flow scheduling problem under a general linear programming (LP) model with core constraints. Third, an effective heuristic algorithm named DCDFS-greedy strategy (DCDFS-GS) is proposed to solve this scheduling problem and reduce the search complexity in multi-domain networks. Finally, simulations in realistic network environments show that DCDFS outperforms the state-of-the-art deterministic schemes for cross-domain scheduling, and optimization algorithms can further improve schedulability and resource utilization.