Scholarlink

<p>Network function virtualization (NFV) has emerged as a transformative new paradigm for enabling service providers to meet increasing connectivity demands while managing operational costs. However, unleashing the full potential of NFV requires the adoption of an effective virtualized network function (VNF) placement algorithm that can dynamically adapt to varying network conditions. At the core of the VNF placement problem is the trade-off between service level agreement (SLA) assurance and energy efficiency, as placing multiple network functions on a single node enables others to be turned off but leads to reduced quality of service. To address this issue, this paper presents combined dynamic predictive auto-scaling (CDPAS) which integrates a neural network and genetic programming model for resource prediction with the combined most full first (CMFF) placement heuristic for optimizing energy consumption. The proposed solution is evaluated against five alternative placement strategies and demonstrates superior performance in jointly reducing energy consumption and SLA violation rates.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Energy-efficient and SLA-aware virtual network function placement using predictive auto-scaling

  • 期刊论文

  • 发表日期：

    2025年12月26日

查看全文
  • Rojia Nikbazm,
  • Mahmood Ahmadi,
  • Mohammad Parsa Toopchinezhad

摘要

Network function virtualization (NFV) has emerged as a transformative new paradigm for enabling service providers to meet increasing connectivity demands while managing operational costs. However, unleashing the full potential of NFV requires the adoption of an effective virtualized network function (VNF) placement algorithm that can dynamically adapt to varying network conditions. At the core of the VNF placement problem is the trade-off between service level agreement (SLA) assurance and energy efficiency, as placing multiple network functions on a single node enables others to be turned off but leads to reduced quality of service. To address this issue, this paper presents combined dynamic predictive auto-scaling (CDPAS) which integrates a neural network and genetic programming model for resource prediction with the combined most full first (CMFF) placement heuristic for optimizing energy consumption. The proposed solution is evaluated against five alternative placement strategies and demonstrates superior performance in jointly reducing energy consumption and SLA violation rates.