Self-Evolving Virtual Network Orchestration From Devices to Cloud Over Shared Infrastructure for Dynamic Services
IEEE Transactions on Mobile Computing · 2026-12 · 25 (12) · 22745-22761
DOI: 10.1109/tmc.2026.3725358 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Transactions on Mobile Computing, 2026-12. Verfasst von Xiying Fan und 6 weiteren Autor:innen. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 36 Literaturangaben. Gefördert von: National Key Research and Development Program, Basic and Applied Basic Research Foundation of Guangdong Province, National Foreign Expert Project.
Abstract (Originalsprache)
End-to-end virtual computing network instance (VCNI) orchestration over shared device-to-cloud infrastructure is essential for supporting heterogeneous services with limited physical resources. However, existing network slicing, virtualization, and edge computing approaches usually focus on partial resource management or fixed orchestration intervals, and often neglect adaptive coordination across the full VCNI lifecycle. Motivated by this, we consider a dynamic device-to-cloud VCNI orchestration scenario that jointly accounts for heterogeneous service demands, wireless spectrum resources, cloud computing resources, and reconfiguration overhead. We propose a self-evolving dual-timescale VCNI orchestration framework to coordinate long-timescale inter-VCNI resource provisioning and short-timescale intra-VCNI user scheduling. The framework adaptively adjusts the VCNI reconfiguration frequency according to workload variations, thereby balancing resource consumption, reconfiguration overhead, and task failure penalties. To tackle this problem, we formulate the orchestration process as a hierarchical Markov decision process (MDP) and design a self-adaptive soft actor-critic algorithm, referred to as SaSAC. SaSAC adopts dual-timescale policies and shared twin critics to learn coordinated orchestration decisions under dynamic service states. Experimental results show that SaSAC achieves lower system cost and higher task success rate than learning-based and heuristic baselines. Moreover, additional ablation and post-training results further validate the effectiveness of SaSAC under varying system conditions and dynamic service demands.
Förderung
- National Key Research and Development Program
- Basic and Applied Basic Research Foundation of Guangdong Province
- National Foreign Expert Project
- Interdisciplinary Research Project for Young Teachers of USTB
Zitieren
APA
Fan, X., Liu, Y., Li, X., Li, C., Huangfu, W., Li, Z., & Zhang, H. (2026). Self-Evolving Virtual Network Orchestration From Devices to Cloud Over Shared Infrastructure for Dynamic Services. IEEE Transactions on Mobile Computing, 25(12), 22745-22761. https://doi.org/10.1109/tmc.2026.3725358
Vancouver
Fan X, Liu Y, Li X, Li C, Huangfu W, Li Z, et al. Self-Evolving Virtual Network Orchestration From Devices to Cloud Over Shared Infrastructure for Dynamic Services. IEEE Transactions on Mobile Computing. 2026;25(12):22745-22761. doi:10.1109/tmc.2026.3725358
BibTeX
@article{Fan2026,
author = {Fan, Xiying and Liu, Yaxi and Li, Xulong and Li, Chengyi and Huangfu, Wei and Li, Zongpeng and Zhang, Haijun},
title = {Self-Evolving Virtual Network Orchestration From Devices to Cloud Over Shared Infrastructure for Dynamic Services},
journal = {IEEE Transactions on Mobile Computing},
year = {2026},
volume = {25},
number = {12},
pages = {22745--22761},
doi = {10.1109/tmc.2026.3725358}
}
EUVIMED