Deadline-Constrained Multi-Agent Collaborative Transmission for Delay-Sensitive Applications

• DOI: 10.1109/tccn.2023.3288133
• Publication Date: 2023-06-23T14:45:58Z
• Authors: Kang Liu; Wei Quan; Nan Cheng; Xue Zhang; Liang Guo; Deyun Gao; Hongke Zhang
• Keywords: Queuing delay

Delay-sensitive applications impose the strict deadline-constrained latency requirement for end-to-end transmission. However, due to high dynamics of link status and limited network resources, it is still challenging ensure deterministic flows. To this end, we propose a multi-agent collaborative transmission mechanism (DMCT). Specially, DMCT featured by combining dynamic routing packet active dropping control avoid additional latency. Firstly, design specific structure detect aware deadline each flow. Secondly, enhance precision decision, further formulate flow scheduling as joint minimization problem that considers flows, queuing delay, status. Then, algorithm find solution problem. Finally, local-agent queue management adjustment. Simulation results show proposed outperforms state-of-the-art solutions, e.g., graph convolutional network-based deep reinforcement learning (GCNL) delay-based shortest path first (D-SPF), in terms one-way round-trip time, drop rate, length.