Hybrid Optimal Load Allocation for Heterogeneous Distributed Computation
IEEE Communications Letters · 2026 · 30 · 3664-3668
DOI: 10.1109/lcomm.2026.3745752 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Communications Letters, 2026. Verfasst von Pengjie Wang und 2 weiteren Autor:innen. Die Arbeit nennt 17 Literaturangaben. Gefördert von: Shaanxi Provincial Key Research and Development Program.
Abstract (Originalsprache)
Current research on heterogeneous distributed computation often suffers from higher latency due to homogeneous processing, or introduces higher encoding/decoding complexity in an attempt to reduce computation latency. To address these issues, this letter proposes a hybrid optimal load allocation (HOLA) scheme, which can further optimize the encoding/decoding complexity of the system while reducing the expected latency. The HOLA scheme innovatively adopts a hybrid fault-tolerance mechanism that integrates passive task reassignment and active coding redundancy, which effectively minimizes system execution time by jointly optimizing the task ratio and the load allocation within groups, and by applying differentiated fault-tolerance strategies to nodes classified according to their reliability. Simulation results show that the proposed HOLA scheme outperforms several existing schemes, with a latency reduction of $\mathbf {31.14}\%$ under typical parameter settings. Furthermore, it reduces encoding and decoding complexity by a factor of approximately $\mathbf {10.9}$ and $\mathbf {3.45}$ , respectively, compared to COLA, while maintaining its latency advantage.
Förderung
- Shaanxi Provincial Key Research and Development Program
Zitieren
APA
Wang, P., Wang, J., & Zhu, Q. (2026). Hybrid Optimal Load Allocation for Heterogeneous Distributed Computation. IEEE Communications Letters, 30, 3664-3668. https://doi.org/10.1109/lcomm.2026.3745752
Vancouver
Wang P, Wang J, Zhu Q. Hybrid Optimal Load Allocation for Heterogeneous Distributed Computation. IEEE Communications Letters. 2026;30:3664-3668. doi:10.1109/lcomm.2026.3745752
BibTeX
@article{Wang2026,
author = {Wang, Pengjie and Wang, Jing and Zhu, Qingzhi},
title = {Hybrid Optimal Load Allocation for Heterogeneous Distributed Computation},
journal = {IEEE Communications Letters},
year = {2026},
volume = {30},
pages = {3664--3668},
doi = {10.1109/lcomm.2026.3745752}
}
EUVIMED