Scholarlink

In this research study, we observe a higher prevalence of Personal Computer (PC) malware families compared to Internet of Things (IoT) malware families. However, the escalating targeting of IoT devices arises from their constrained resources and inadequate security measures. Developing effective security for IoT devices necessitates substantial computational power, which they lack. Furthermore, constant Internet connectivity of IoT devices elevates the susceptibility to attacks. Unlike PCs, IoT platforms exhibit greater diversity, complicating detection and requiring specialized tools. Thus, the detection of malware in IoT networks holds utmost significance, attracting significant research attention. Our study demonstrates the strong performance of LSTM and Bi-LSTM with accuracy rates of 99.91% and 99.98%, respectively. The classification process highlights critical features influencing outcomes, including timing, historical data, protocols used, and transmitted origin bytes. Prominently, identifiable trends pertaining to operations such as port scanning, infections, and distributed denial of service (DDoS) assaults show up time and volume-wise from particular IP addresses constantly.

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

Detecting Malware in IoT Networks Using Advanced Learning Techniques

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Rajaram Hansda,
  • Jyoti Grover,
  • Sunita Singhal,
  • Vanisha Kheterpal

摘要

In this research study, we observe a higher prevalence of Personal Computer (PC) malware families compared to Internet of Things (IoT) malware families. However, the escalating targeting of IoT devices arises from their constrained resources and inadequate security measures. Developing effective security for IoT devices necessitates substantial computational power, which they lack. Furthermore, constant Internet connectivity of IoT devices elevates the susceptibility to attacks. Unlike PCs, IoT platforms exhibit greater diversity, complicating detection and requiring specialized tools. Thus, the detection of malware in IoT networks holds utmost significance, attracting significant research attention. Our study demonstrates the strong performance of LSTM and Bi-LSTM with accuracy rates of 99.91% and 99.98%, respectively. The classification process highlights critical features influencing outcomes, including timing, historical data, protocols used, and transmitted origin bytes. Prominently, identifiable trends pertaining to operations such as port scanning, infections, and distributed denial of service (DDoS) assaults show up time and volume-wise from particular IP addresses constantly.