Research Article
Transfer Learning Based Algorithm for Service Deployment Under Microservice Architecture
@INPROCEEDINGS{10.1007/978-3-030-99200-2_5, author={Wenlin Li and Bei Liu and Hui Gao and Xin Su}, title={Transfer Learning Based Algorithm for Service Deployment Under Microservice Architecture}, proceedings={Communications and Networking. 16th EAI International Conference, ChinaCom 2021, Virtual Event, November 21-22, 2021, Proceedings}, proceedings_a={CHINACOM}, year={2022}, month={4}, keywords={Service deployment Edge computing DQN Transfer learning}, doi={10.1007/978-3-030-99200-2_5} }- Wenlin Li
Bei Liu
Hui Gao
Xin Su
Year: 2022
Transfer Learning Based Algorithm for Service Deployment Under Microservice Architecture
CHINACOM
Springer
DOI: 10.1007/978-3-030-99200-2_5
Abstract
In recent years, with the large-scale deployment of 5G network, research on 6G networks has gradually begun. In the 6G era, new service scenarios, such as Broad Coverage and High Latency Communication (BCHLC) will be introduced into the network, further increasing the complexity of network management. Furthermore, the development of edge computing and microservice architectures enables services to be deployed in a container on the edge clouds closer to the user side, significantly solving the problems. However, how to deploy services on edge clouds with limited resources is still an unresolved problem. In this paper, we model the problem as a Markov Decision Process (MDP), then propose a Deep Q Learning (DQN) based service deployment algorithm to optimize the delay and deployment cost of the services. Furthermore, a Multi-Category Joint Optimization Transfer Learning (MCJOTL) algorithm is proposed in this paper to address the problem of slow convergence of the DQN algorithm, which can adapt to different service scenarios in future networks faster. The simulation results show that the proposed algorithm can effectively improve training efficiency and service deployment effects.
