Scholarlink

<p>The Internet of Things (IoT) is gaining substantial popularity with the advancement of smart technologies. IoT applications have expanded due to exponential growth in smart devices and decreasing sensor costs. However, IoT devices are vulnerable to a variety of security attacks. Intrusion Detection System (IDS) serve an essential role in detecting and responding to these attacks by monitoring device behavior, traffic, and system logs for indications of malicious activity. Traditional IDS solutions face challenges such as high data complexity, difficulty in accurately detecting attacks, security issues, and low detection accuracy. To address these challenges, this study proposes a novel Deep Learning (DL)-based IDS integrated&#xa0;with a lattice-based cryptography algorithm&#xa0;to ensure secure data transfer in IoT environments. Initially, the intrusion dataset is collected&#xa0;and a preprocessing step is carried out to enhance its quality by eliminating unnecessary data and normalizing the dataset. Stacked Sparse Autoencoder (SSAE) is utilized to extract high-level features from the preprocessed dataset. Following, the extracted&#xa0;features are fed into the proposed&#xa0;Hybrid Peep-GhostNet (HPGNet), which combines GhostNet and Peephole LSTM (P-LSTM) to accurately classify data as normal or malicious.&#xa0;Next, normal data is encrypted utilizing the Enhanced Ring Learning with Errors (ERing-LWE) algorithm&#xa0;to ensure secure data transmission. To further improve security with slower computation, the Improved Mother Optimization (IMO) algorithm is used to optimize the parameters of the Ring-LWE algorithm. Experimental validation on the NSL-KDD dataset shows that the proposed framework obtains a maximum accuracy of 99.82% and security level of 98.79% compared to current techniques and provides enhanced security for IoT devices during data transmission.</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

Enhancing IoT network security with hybrid Peep-GhostNet and enhanced ring learning with errors cryptography algorithm

  • 期刊论文

  • 发表日期：

    2025年9月22日

查看全文
  • K. Swathi,
  • G. Hima Bindu

摘要

The Internet of Things (IoT) is gaining substantial popularity with the advancement of smart technologies. IoT applications have expanded due to exponential growth in smart devices and decreasing sensor costs. However, IoT devices are vulnerable to a variety of security attacks. Intrusion Detection System (IDS) serve an essential role in detecting and responding to these attacks by monitoring device behavior, traffic, and system logs for indications of malicious activity. Traditional IDS solutions face challenges such as high data complexity, difficulty in accurately detecting attacks, security issues, and low detection accuracy. To address these challenges, this study proposes a novel Deep Learning (DL)-based IDS integrated with a lattice-based cryptography algorithm to ensure secure data transfer in IoT environments. Initially, the intrusion dataset is collected and a preprocessing step is carried out to enhance its quality by eliminating unnecessary data and normalizing the dataset. Stacked Sparse Autoencoder (SSAE) is utilized to extract high-level features from the preprocessed dataset. Following, the extracted features are fed into the proposed Hybrid Peep-GhostNet (HPGNet), which combines GhostNet and Peephole LSTM (P-LSTM) to accurately classify data as normal or malicious. Next, normal data is encrypted utilizing the Enhanced Ring Learning with Errors (ERing-LWE) algorithm to ensure secure data transmission. To further improve security with slower computation, the Improved Mother Optimization (IMO) algorithm is used to optimize the parameters of the Ring-LWE algorithm. Experimental validation on the NSL-KDD dataset shows that the proposed framework obtains a maximum accuracy of 99.82% and security level of 98.79% compared to current techniques and provides enhanced security for IoT devices during data transmission.