Packet-Level Adaptive Forwarding with Multi-Cyclic Queues for Time-Sensitive Networks Using Delay Slack

Abstract

Time-sensitive networking (TSN) aims to provide bounded latency and reliable delivery for critical applications. Existing cyclic queuing and forwarding (CQF)-based mechanisms commonly rely on centralized flow-level scheduling, path computation, and resource reservation, requiring prior knowledge of traffic demands and the global network state. Although effective for predictable time-triggered flows, such mechanisms may suffer from limited scalability and slow adaptation to online or bursty traffic. This paper proposes a packet-level adaptive forwarding and multi-cyclic queue scheduling mechanism for CQF-based time-sensitive networks. The proposed mechanism enables online forwarding and queue selection at intermediate nodes by combining a packet-carried forwarding state with local queue-state information. A slack-aware capacity-balancing enqueue rule is designed over the multi-cyclic queue model to allocate queue resources according to packet delay slack and queue availability. Simulation results show that, compared with representative CQF-based baselines, the proposed mechanism reduces the service completion time, improves delivery success and service capacity under bursty or locally congested conditions, and maintains a better queue load balance. These results suggest a limited but practical role for the proposed mechanism. It can be used as a packet-level forwarding scheme, and it can also complement centralized CQF-based scheduling when traffic changes too quickly for full prescheduling.