Failure Prediction Based VNF Migration Mechanism for Multimedia Services in Power Grid Substation Monitoring

Wen, Mingshi; Li, Junqin; Hai, Tianxiang; Wang, Pingping; Yin, Kang; Hao, Jiakai; Diwen, Zhu; Yang, Yang; Tong, Riming

doi:10.1109/bmsb55706.2022.9828707

Cited by 2 publications

(1 citation statement)

References 5 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, only a limited number of studies consider VNF failure and post-failure recovery mechanisms from the perspective of network management, such as, intelligent and timely VNF failure detection [2], VNF backup assignment as a protection strategy [9], faulty SFC repositioning and [9]. Another research study proposed a deep reinforcement learning model to predict VNF failure and migrate affected ones to non-repudiation hosts while simultaneously maintaining QoS [10]. Likewise, Avgeris et al attempt to address SFC failure by repositioning VNFs into successfully running data centers by using a resourcedriven reinforcement learning mechanism based on Stochastic Learning Automata [9].…”

Section: Related Workmentioning

confidence: 99%

Robust Virtual Network Function Optimization Under Post-Failure Uncertainty: Classical & Quantum Computi

Emu,

Choudhury,

Salomaa

2024

Preprint

View full text Add to dashboard Cite

The Service Function Chaining (SFC) failure is a rare event that comes with escalating unexpected costs. It is an overly simplistic assumption that during the installation of the Virtual Network Function (VNF) recovery instance, the network conditions impacting the cost of redeployment will remain unchanged. In this paper, we propose a deterministic optimization model using traditional Integer Linear Programming (ILP) that maneuvers the resource allocation for prior and post-failure SFC deployment. Afterwards, we design a robust optimization model that accounts for the uncertainty of the redeployment costs. As per the strong duality theorem, we derive the dual formulation of the robust optimization model for reduced computational complexity. Further along this line, we propose a quantum annealing-driven quadratic optimization (QUBO) model that demonstrates inherent robustness even without explicitly considering the uncertainty bounds of SFC redeployment costs. Extensive simulation studies demonstrate the superiority of robust solutions over deterministic approaches and explore the potential strengths of quantum annealing in terms of intrinsic resiliency. Although quantum computing is not yet ready to solve large-scale SFC deployment, it can support VNF services that demand ultra-low running time/real-time decision-making.

show abstract

Section: Related Workmentioning

confidence: 99%