Scholarlink

<p>Securing data communication in Internet of Things (IoT) networks is challenging due to limited power, processing capacity, and the heterogeneous nature of devices. Traditional encryption methods are often resource-intensive, while conventional authentication schemes lack adaptability, limiting their applicability in mission-critical domains such as healthcare, smart surveillance, and autonomous systems. This study presents a unified security framework that integrates lightweight Hybrid Encryption Algorithms (HEA) with machine learning–based authentication to enhance secure communication across diverse sensor nodes. A real-time testbed comprising temperature sensors, ECG modules, and Raspberry Pi-based image sensors was developed for evaluation. The proposed hybrid encryption scheme, optimized for low-power, low-memory IoT devices, outperforms AES and RSA in speed and efficiency, while a novel S-box design improves non-linearity and cryptanalysis resistance. A machine learning-driven authentication module enables real-time anomaly detection and access control, adapting dynamically to device behaviour patterns. Comparative analysis against benchmark algorithms, including PRESENT, SIMON, SPECK, TEA, and a custom S-box-based approach, demonstrates lower latency, reduced computational overhead, and improved scalability. The results validate the framework’s practicality for secure, low-latency communication in next-generation IoT 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

A hybrid framework for secure IoT communication using lightweight cryptography and machine learning-based authentication

  • 期刊论文

  • 发表日期：

    2025年10月23日

查看全文
  • Geeta Sharma,
  • Narinder Verma,
  • Sanjeev Jain,
  • Radhe Shyam Sharma

摘要

Securing data communication in Internet of Things (IoT) networks is challenging due to limited power, processing capacity, and the heterogeneous nature of devices. Traditional encryption methods are often resource-intensive, while conventional authentication schemes lack adaptability, limiting their applicability in mission-critical domains such as healthcare, smart surveillance, and autonomous systems. This study presents a unified security framework that integrates lightweight Hybrid Encryption Algorithms (HEA) with machine learning–based authentication to enhance secure communication across diverse sensor nodes. A real-time testbed comprising temperature sensors, ECG modules, and Raspberry Pi-based image sensors was developed for evaluation. The proposed hybrid encryption scheme, optimized for low-power, low-memory IoT devices, outperforms AES and RSA in speed and efficiency, while a novel S-box design improves non-linearity and cryptanalysis resistance. A machine learning-driven authentication module enables real-time anomaly detection and access control, adapting dynamically to device behaviour patterns. Comparative analysis against benchmark algorithms, including PRESENT, SIMON, SPECK, TEA, and a custom S-box-based approach, demonstrates lower latency, reduced computational overhead, and improved scalability. The results validate the framework’s practicality for secure, low-latency communication in next-generation IoT applications.