Scholarlink

<p>Within the constrained resources, intrusions undermine the availability and reliability of wireless sensor networks and therefore proper and effective detection is necessary. This paper is introducing the intrusion detection system, GWORNN, based on grey-wolf optimization of selecting features and a recurrent network classifier operating in a sequence. We test three population sizes, namely 3, 5 and 7 wolves, with the aim to examine exploration-exploitation trade-off of optimal optimizer. The 7wolf GWORNN achieves accuracy of 96% on the NSL KDD data, which is a high score compared to the 5wolf version with 92%, the RNN baseline with 89% and even the 3wolf version with 79%. The 7-wolf structure also has a 96 per cent detection rate and is able to isolate a small feature set which improves the efficiency of deploying WSNs. The results reveal that GWORNN enhances the performance of detection and lowers input dimensionality and inference time in comparison to the baseline IDS techniques.</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

Advanced GWORNN based Intrusion Detection Scheme for Securing Wireless Sensor Networks

  • 期刊论文

  • 发表日期：

    2026年6月20日

查看全文
  • Karthic Sundaram,
  • S. Manoj Kumar

摘要

Within the constrained resources, intrusions undermine the availability and reliability of wireless sensor networks and therefore proper and effective detection is necessary. This paper is introducing the intrusion detection system, GWORNN, based on grey-wolf optimization of selecting features and a recurrent network classifier operating in a sequence. We test three population sizes, namely 3, 5 and 7 wolves, with the aim to examine exploration-exploitation trade-off of optimal optimizer. The 7wolf GWORNN achieves accuracy of 96% on the NSL KDD data, which is a high score compared to the 5wolf version with 92%, the RNN baseline with 89% and even the 3wolf version with 79%. The 7-wolf structure also has a 96 per cent detection rate and is able to isolate a small feature set which improves the efficiency of deploying WSNs. The results reveal that GWORNN enhances the performance of detection and lowers input dimensionality and inference time in comparison to the baseline IDS techniques.