Scholarlink

<p>The increasing complexity and heterogeneity of modern network traffic present significant challenges for accurate traffic classification in Software-Defined Networking (SDN) environments. Conventional classification methods repetitively struggle to comply to expeditiously evolving traffic behaviors, resulting in suboptimal resource management. To address this issue, an enhanced Extreme Learning Machine with Ridge Regression (RR-ELM) is proposed for efficient and accurate SDN traffic classification. The model is assessed on the ISCX VPN non – VPN dataset, which contains 28 flow-level features across eight application categories such as audio, video, and browsing etc. The incorporation of ridge regression improves the stability and generalization of the standard ELM, making it suitable for dynamic SDN environments. Experimental results determine that the proposed RR-ELM model accomplishes a classification accuracy of 94.35%, surpassing both Convolutional Neural Networks and standard ELM. These findings show that RR-ELM offers a lightweight yet effective solution for real-time traffic classification in SDN.</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

An Enhanced Extreme Learning Machine With Ridge Regression for Network Traffic Classification in Software-defined Networking

  • 期刊论文

  • 发表日期：

    2026年2月6日

查看全文
  • U. Prabu,
  • V. Geetha

摘要

The increasing complexity and heterogeneity of modern network traffic present significant challenges for accurate traffic classification in Software-Defined Networking (SDN) environments. Conventional classification methods repetitively struggle to comply to expeditiously evolving traffic behaviors, resulting in suboptimal resource management. To address this issue, an enhanced Extreme Learning Machine with Ridge Regression (RR-ELM) is proposed for efficient and accurate SDN traffic classification. The model is assessed on the ISCX VPN non – VPN dataset, which contains 28 flow-level features across eight application categories such as audio, video, and browsing etc. The incorporation of ridge regression improves the stability and generalization of the standard ELM, making it suitable for dynamic SDN environments. Experimental results determine that the proposed RR-ELM model accomplishes a classification accuracy of 94.35%, surpassing both Convolutional Neural Networks and standard ELM. These findings show that RR-ELM offers a lightweight yet effective solution for real-time traffic classification in SDN.