Achieving Class-Aware Mixed-Flow Scheduling in Hybrid Wired-Wireless Time-Sensitive Networks

The emergence of new time-sensitive networking (TSN) technologies empowers almost-deterministic ultra-reliable low-latency communications for industrial cloud-fog automation paradigms. However, current heterogeneous networks struggle to balance time-sensitivity and flexibility, particularly in mixed-flow scenarios. Existing scheduling approaches within 5G-TSN integration model either suffer from quality of service (QoS) flow mismatch or microbursts. This paper proposes a class-aware mixed-flow scheduling (CAMFS) strategy to enable domain-specific resource allocation in a hybrid wired-wireless TSN, while meeting the differentiated time-sensitive (TS) requirements. A time-triggered multiple cyclic-queuing (TTMCQ) shaper is designed to effectively align the classified and regulated mixed flows from TSN domains to 5G QoS flows through class-aware mapping. We also introduce an anti-starvation resource optimization method that minimizes the total average idle resources resulting from temporal-spatial resource over-provisioning within reserved TS windows (RTWs). Additionally, we present a CAMFS algorithm aimed at enhancing schedulability and resource utilization by sorting mixed flows based on a combination of flow features. Finally, simulation results show that CAMFS exhibits outstanding scheduling performance in terms of end-to-end service latency, scheduling success ratio, and normalized resource distribution compared to other state-of-the-art methods.