BROWSE
Detail
Conference Paper
CollabOffloading: A Computational Offloading Methodology Using External Clouds for Limited Private On-Site Edge Servers
Cited 0 time in 
Share 
Share
- Authors
- Junhee Lee, Jaeho Jeon, SungJoo Kang
- Issue Date
- 2021-12
- Citation
- Asia Simulation Conference (AsiaSim) 2021 (CCIS 1636), pp.179-186
- Publisher
- Springer
- Language
- English
- Type
- Conference Paper
- Abstract
- In this paper, we proposed a methodology using Kubernetes clustered on-site edge servers with external clouds to provide computational offloading functionality for resource-limited private edge servers. This methodology enables additional functionalities without changing hardware infrastructures for industrial areas such as manufacturing systems. We devised a compute-intensive task scheduling algorithm using real-time CPU usage information of Kubernetes cluster to determine computation offloading decision. The purpose of the experiment is to compare overall performance between on-site edge only cluster and external cloud offloading cluster. The experiment scenario contains complex simulation problem which selects optimal tollgate for congested traffic situation. The result of experiment shows the proposed CollabOffloading methodology reduces entire execution time of simulations.
- KSP Keywords
- CPU usage, Computation offloading, Computational offloading, External cloud, Manufacturing system, On-site, Overall performance, Real-time, Task scheduling algorithm, Usage information, cloud offloading