Scholarlink

<p>The proliferation of Internet of Things (IoT) devices introduces severe cybersecurity challenges that conventional Intrusion Detection Systems (IDS) cannot adequately resolve due to their computational inefficiency and high resource demands. This paper presents a novel lightweight IDS framework leveraging a Hybrid Genetic Algorithm and Binary Grey Wolf Optimizer (HGABGWO) for feature selection, integrated with Support Vector Machine (SVM) classification. The proposed hybrid model incorporates a multi-objective weighted fitness function to optimize feature subsets, effectively balancing detection accuracy and computational overhead by enhancing search capability and convergence speed. Extensive evaluations on the NSL-KDD and AWID datasets demonstrate superior performance, achieving 99.73% accuracy with only 20 features and 0.07% false positive rate on NSL-KDD, and 99.06% accuracy with 89 features on AWID. These results consistently surpass both baseline algorithms (GA and BGWO) and recent state-of-the-art methods, confirming the robustness and efficiency of the proposed system for real-world IoT and TCP/IP network security applications.</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 efficient intrusion detection system for IoT and TCP/IP networks based on hybrid genetic and binary Grey Wolf Algorithms with a multi-objective fitness function

  • 期刊论文

  • 发表日期：

    2026年1月6日

查看全文
  • Sheyda Davahli

摘要

The proliferation of Internet of Things (IoT) devices introduces severe cybersecurity challenges that conventional Intrusion Detection Systems (IDS) cannot adequately resolve due to their computational inefficiency and high resource demands. This paper presents a novel lightweight IDS framework leveraging a Hybrid Genetic Algorithm and Binary Grey Wolf Optimizer (HGABGWO) for feature selection, integrated with Support Vector Machine (SVM) classification. The proposed hybrid model incorporates a multi-objective weighted fitness function to optimize feature subsets, effectively balancing detection accuracy and computational overhead by enhancing search capability and convergence speed. Extensive evaluations on the NSL-KDD and AWID datasets demonstrate superior performance, achieving 99.73% accuracy with only 20 features and 0.07% false positive rate on NSL-KDD, and 99.06% accuracy with 89 features on AWID. These results consistently surpass both baseline algorithms (GA and BGWO) and recent state-of-the-art methods, confirming the robustness and efficiency of the proposed system for real-world IoT and TCP/IP network security applications.