Time-Efficient Task Offloading and Decentralized Collaborative Scheduling Method for Cross-Domain Computing Power Networks
Zhao, Dongcheng Xu, Guangxia Sun, Gang Guizani, Mohsen
Zhao, Dongcheng
Xu, Guangxia
Sun, Gang
Guizani, Mohsen
Supervisor
Department
Machine Learning
Embargo End Date
Type
Journal article
Date
License
Language
English
Collections
Research Projects
Organizational Units
Journal Issue
Abstract
The combination of task offloading and resource scheduling of computing power networks forms a distributed computing paradigm. Meanwhile, the integration of computing power networks with the C2F (Customer-to-Factory) model and task offloading technology is reshaping the relationship between consumption and production. However, most current studies on task offloading are based on centralized cloud/edge networks and are not applicable to large-scale distributed cross-domain computing power networks. Therefore, this paper studied task offloading based on cross-domain computing power networks, thereby reducing the scheduling time of the algorithm. Thereby, this work proposed the decentralized collaborative task offloading algorithm based on fast layering (DCTO-FL). Through a fast layering strategy, each domain obtains a layered set of backup domains, and then collaborative task offloading is carried out. When the user’s domain cannot meet the resource or delay requirements of the task, the management node of the user’s domain will publish the task to the management nodes of the backup domains, thereby achieving rapid cross-domain collaborative deployment. Finally, we verified that the DCTO-FL algorithm is advantageous, which task offloading cost is at most 10% lower, running time is at most 95% lower, and task delay with running time is at most 90% than the baseline.
Citation
D. Zhao, G. Xu, G. Sun, M. Guizani, "Time-Efficient Task Offloading and Decentralized Collaborative Scheduling Method for Cross-Domain Computing Power Networks," Computer Networks, pp. 112294-112294, 2026, https://doi.org/10.1016/j.comnet.2026.112294.
Source
Computer Networks
Conference
Keywords
46 Information and Computing Sciences, 4605 Data Management and Data Science, 4606 Distributed Computing and Systems Software
Subjects
Source
Publisher
Elsevier