Scholarlink

With the significant growth of global Internet of Things (IoT) devices and traffic, the demand of seamless global coverage services is increasing. Low Earth Orbit (LEO) satellites are an excellent solution. However, the dynamically changing topology makes it difficult to provide stable delay-sensitive services for industrial IoT, meanwhile, the limited on-board resources greatly reduce the stability and reliability of the services. In order to achieve deterministic transmission for time-sensitive services on LEO satellites, this paper constructs a LEO time-sensitive network architecture. In this architecture, we establish a satellite network port time slot scheduling model to avoid transmission conflicts of time-sensitive flows. Furthermore, we utilize a time slot state evaluation process to optimize the scheduling model. To solve the optimization problem, we propose an improved Deep-Q Network (DQN) algorithm. The algorithm uses breadth first search (BFS) to search out the feasible path set for each flow, and selects the conflict-free path set for all flows by DQN algorithm. The experimental results show compared with other strategies, the proposed strategy reduces the average delay of flows and improves the amount of available time slots for best-effort (BE) flows.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A Flow Scheduling Strategy Based on the DQN Algorithm for Low Earth Orbit Satellite Time-Sensitive Networks

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Fei Zheng,
  • Juzhong Wei,
  • Longjie Chen,
  • Dongyu Li,
  • Jilong Zhao,
  • Yu Hu

摘要

With the significant growth of global Internet of Things (IoT) devices and traffic, the demand of seamless global coverage services is increasing. Low Earth Orbit (LEO) satellites are an excellent solution. However, the dynamically changing topology makes it difficult to provide stable delay-sensitive services for industrial IoT, meanwhile, the limited on-board resources greatly reduce the stability and reliability of the services. In order to achieve deterministic transmission for time-sensitive services on LEO satellites, this paper constructs a LEO time-sensitive network architecture. In this architecture, we establish a satellite network port time slot scheduling model to avoid transmission conflicts of time-sensitive flows. Furthermore, we utilize a time slot state evaluation process to optimize the scheduling model. To solve the optimization problem, we propose an improved Deep-Q Network (DQN) algorithm. The algorithm uses breadth first search (BFS) to search out the feasible path set for each flow, and selects the conflict-free path set for all flows by DQN algorithm. The experimental results show compared with other strategies, the proposed strategy reduces the average delay of flows and improves the amount of available time slots for best-effort (BE) flows.