Adaptive Packet Scheduling Algorithm for Time-Sensitive Service based on DTP and Reinforcement Learning

With the rapid development of mobile Internet and Internet of things, a series of time-sensitive services such as video conferencing, cloud games, AR / VR have emerged. In order to meet the above time-sensitive services, IETF has proposed Deterministic Network (DetNet) architecture, which provides an ideal deterministic delay through clock synchronization, zero congestion loss and other mechanisms. Deadline-aware Transport Protocol (DTP) is design to specify the deadline in application layer, and then meets the requirements in transport layer, which can support the time-sensitive services easily. However, the current packet transmission mechanisms are all rule-based and relatively static strategies, which cannot meet the demand of time-sensitive service in dynamic networks. Therefore, this paper proposes an algorithm to dynamically adjust the transmission priority by combining the DTP and reinforcement learning to tackle this issue. More specifically, we design the reward function according to the requirements of DTP, and propose the algorithm of congestion control and packet scheduling in the transport layer. We consider not only the priority but also the service deadline. Comprehensive experiments show that our algorithm performs better in the transmission of time-sensitive services compared to traditional packet scheduling strategies.