Distributed Model Training based on Data Parallelism in Edge Computing-enabled Elastic Optical Networks
Journal article, 2021
IEEE The emergence of edge computing provides an effective solution to execute distributed model training (DMT). The deployment of training data among edge nodes affects the training efficiency and network resource usage. This letter aims for the efficient provisioning of DMT services by optimizing the partition and distribution of training data in edge computing-enabled optical networks. An integer linear programming (ILP) model and a data parallelism deployment algorithm (DPDA) are proposed to solve this problem. The performance of the proposed approaches is evaluated through simulation. Simulation results show that the proposed algorithm can deploy more DMT services compared with benchmark.
edge computing
Parallel processing
optical networks
distributed model training
Training
Task analysis
Data models
Data parallelism
Training data
Computational modeling
Optical fiber networks
Author
Yajie Li
Beijing University of Posts and Telecommunications (BUPT)
Zebin Zeng
Beijing University of Posts and Telecommunications (BUPT)
Jun Li
Chalmers, Electrical Engineering, Communication, Antennas and Optical Networks
Boyuan Yan
Beijing University of Posts and Telecommunications (BUPT)
Yongli Zhao
Beijing University of Posts and Telecommunications (BUPT)
Jie Zhang
Beijing University of Posts and Telecommunications (BUPT)
IEEE Communications Letters
1089-7798 (ISSN) 15582558 (eISSN)
Vol. 25 4 1241-1244 9274363Subject Categories
Computer Engineering
Communication Systems
Computer Science
DOI
10.1109/LCOMM.2020.3041453