RSDD: Resilient Statistical Drift Detection for vSDN Hypervisors
IEEE Networking Letters · 2026 · 8 · 301-305
DOI: 10.1109/lnet.2026.3683210 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Networking Letters, 2026. Verfasst von Mohamed Khalafalla Hassan und 4 weiteren Autor:innen. Die Arbeit nennt 16 Literaturangaben. Gefördert von: Taighde Éireann-Research Ireland, U.S.–Ireland Research and Development Partnership Program Project “Resilient Networks”.
Abstract (Originalsprache)
Sixth-Generation (6G) virtualized Software-Defined Networks (vSDNs) require hypervisors that remain stable under non-stationary and bursty traffic. Existing systems, including DLVisor, still rely on a single Anderson–Darling test, which reacts poorly to gradual or variance-driven distributional shifts. This work introduces a unified drift-aware hypervisor, the Resilient Statistical Drift Detection (RSDD) framework, combining KSWIN, Page–Hinkley, Jensen–Shannon divergence, and variance monitoring into a consensus drift-severity index that guides an xLSTM forecaster and cost-aware slice reallocator. Experiments on Tier-2 ISP traces show RSDD significantly reduces overutilization events while maintaining SLA-compliant bandwidth allocation.
Förderung
- Taighde Éireann-Research Ireland
- U.S.–Ireland Research and Development Partnership Program Project “Resilient Networks”
Zitieren
APA
Hassan, M. K., Cherkaoui, I., Alshareef, I. Y., Hamam, H., & Dey, I. (2026). RSDD: Resilient Statistical Drift Detection for vSDN Hypervisors. IEEE Networking Letters, 8, 301-305. https://doi.org/10.1109/lnet.2026.3683210
Vancouver
Hassan MK, Cherkaoui I, Alshareef IY, Hamam H, Dey I. RSDD: Resilient Statistical Drift Detection for vSDN Hypervisors. IEEE Networking Letters. 2026;8:301-305. doi:10.1109/lnet.2026.3683210
BibTeX
@article{Hassan2026,
author = {Hassan, Mohamed Khalafalla and Cherkaoui, Ilias and Alshareef, Ibrahim Y. and Hamam, Habib and Dey, Indrakshi},
title = {RSDD: Resilient Statistical Drift Detection for vSDN Hypervisors},
journal = {IEEE Networking Letters},
year = {2026},
volume = {8},
pages = {301--305},
doi = {10.1109/lnet.2026.3683210}
}
EUVIMED