Time sensitive network traffic scheduling system and method based on causal data completion

By employing a deep reinforcement learning approach that combines causal data completion and hierarchical optimization, the problems of decreased decision-making performance and unmet multi-service requirements of the DRL scheduling algorithm under non-ideal observation conditions are solved, achieving deterministic assurance of critical traffic and efficient transmission of ordinary traffic.

CN122093330APending Publication Date: 2026-05-26TIANJIN UNIV
0 Cites 0 Cited by

Patent Information

Application Number
CN202610174619.7
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-02-06
Publication Date
2026-05-26

AI Technical Summary

Technical Problem

Existing DRL scheduling algorithms cannot accurately perceive the environmental state under non-ideal observation conditions, resulting in a decline in decision-making performance. Furthermore, a single optimization objective cannot meet the differentiated needs of multiple services, especially the lack of deterministic guarantee for critical traffic.

Method used

A TSN traffic scheduling method based on causal data completion is adopted. The missing data is completed by a neural Granger causal state reconstruction module, robust spatiotemporal features are extracted by a variational graph attention network, and a hierarchical optimized deep reinforcement learning decision network is used to optimize the scheduling strategy of critical traffic and ordinary traffic.

Benefits of technology

The availability of the scheduler and network reliability were improved under non-ideal observation conditions, ensuring hard real-time performance of critical traffic while improving the transmission efficiency and throughput of ordinary traffic.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The invention belongs to the technical field of communication and network control, and aims to solve the problems that the adaptability is poor in a non-ideal observation environment and a single optimization target cannot meet various differentiated service requirements, and overcome scheduling failure when data is lost due to network jitter or node failure. According to the time-sensitive network traffic scheduling method based on causal data completion, the real-time traffic of the current time-sensitive TSN is collected, and an irregular observation matrix is constructed; complementing missing data in the observation matrix by using a state reconstruction module based on neural Granger causality; extracting robust spatial-temporal features with anti-noise capability in the complemented observation matrix through a variational graph attention network; and inputting the robust spatial-temporal characteristics with the anti-noise capability into a reinforcement learning decision network based on hierarchical optimization, optimizing the best effort BE stream transmission efficiency under the real-time constraint of a priority key time triggered TT stream hard, and finally generating a gating control list to be issued and executed. The method is mainly applied to design and manufacturing occasions.
Need to check novelty before this filing date? Find Prior Art