Journal of Grid Computing Программное обеспечение

Edge Server Layout and Task Offloading in Dynamic Environment Based on Grey Numbers

Hao, Yongsheng, Wang, Kunfu, Li, Lei, Cao, Jie
Springer Nature · 2026-11-10 · DOI 10.1007/s10723-026-09840-x

Аннотация

The deployment of edge servers and the offloading of tasks have gained much attention in the cloud-edge-end environment. Previous studies have largely neglected three critical uncertainties: the dynamic nature of energy harvesting at edge nodes, the variability in network conditions among edge nodes, and the fluctuations in task scale, which include variations in input/output file sizes and instruction counts, and the bandwidth. These uncertainties have been acknowledged in the literature, underscoring their importance in the optimization of edge-computing frameworks. To address these challenges, we use a grey number comparison approach to evaluate edge server deployment schemes whose energy, bandwidth, and task-size parameters are only known as bounded uncertain quantities, and then select the most suitable scheme. To reduce the complexity of the comparisons of all optimal edge server deployment schemes, we leverage a grey wolf optimization algorithm to identify the most effective scheduling scheme. Besides that, a task scheduling based on grey numbers is also proposed after the edge server has been developed. Our comparative analysis reveals that the proposed grey number-based grey wolf algorithm significantly outperforms the existing methods in terms of reducing both energy consumption and execution time. This work not only deepens the understanding of bounded uncertainties in edge computing, but also offers a robust framework for enhancing task migration and deployment strategies in dynamic environments.

Читать у издателя →