DAG micro-service migration method and device based on deep reinforcement learning

Through the action execution layered reinforcement learning microservice migration algorithm based on deep reinforcement learning, the migration target is decomposed using high-level and low-level actions, the problem of complex dependencies of DAG tasks and the inability to obtain global information in the edge computing system is solved, and efficient microservice migration decisions and better service experience are achieved.

CN119997099AActive Publication Date: 2025-05-13HARBIN INSTITUTE OF TECHNOLOGY (SHENZHEN) (INSTITUTE OF SCIENCE AND TECHNOLOGY INNOVATION HARBIN INSTITUTE OF TECHNOLOGY SHENZHEN)
View PDF 4 Cites 1 Cited by

Patent Information

Application Number
CN202411940273.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-12-26
Publication Date
2025-05-13
Estimated Expiration
2044-12-26

AI Technical Summary

Technical Problem

The prior art is difficult to effectively handle complex dependencies in DAG tasks, and global information cannot be obtained in real time in edge computing systems, resulting in difficulty in making decisions on microservice migration.

Method used

The action execution layered reinforcement learning microservice migration algorithm based on deep reinforcement learning is adopted to decompose migration targets through high-level and low-level actions, use local information to make microservice migration decisions, and process long-term dependency information in the LSTM network.

Benefits of technology

It effectively solves the problem of DAG microservice migration in edge environments with unknown global information, reduces the action space of reinforcement learning, improves model training efficiency, and provides a better service experience in the Internet of Vehicles scenario.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119997099A_ABST
    Figure CN119997099A_ABST
Patent Text Reader

Abstract

The invention discloses a DAG micro-service migration method and device based on deep reinforcement learning, and the method comprises the steps: carrying out the training of a micro-service migration model through employing a reinforcement learning micro-service migration algorithm based on motion execution layering, and obtaining a trained micro-service migration model; deploying the trained micro-service migration model to a user side, and when the position of the user side moves, making a migration decision of the micro-service by the trained micro-service migration model according to the obtained local information to obtain a target edge server; and executing migration of the micro service based on the target edge server. Through the hierarchical decision design, the problem of DAG micro-service migration in an edge environment with unknown global information, especially in an Internet of Vehicles scene, can be effectively solved.
Need to check novelty before this filing date? Find Prior Art

Citation Information

Patent Citations

  • DNN partitioning method, edge computing architecture adopting same and storage medium

    CN114281535A

  • Edge computing task unloading method based on assistance of high-altitude communication platform

    CN116489668A

  • Service function deployment and unmanned aerial vehicle edge node resource allocation method based on deep reinforcement learning

    CN118784650A

  • Agent policy learning method with privacy protection in mobile edge computing

    WO2024254892A1

Cited By

  • Thermal migration method and device for power distribution network service, electronic equipment, storage medium and computer program product

    CN121029435A