Hybrid Fusion of Lightweight Security Frameworks Using Data Mining Approach in IoT
Abhishek Kumar,Samta Jain ..,Sumit Kumar,Hitesh Kumar Sharma,,,,
DOI: https://doi.org/10.54216/fpa.140220
2024-01-01
Abstract:The rapid adoption of the Internet of Things throughout healthcare and smart city construction has led to a rise in networked devices and security issues. This work suggests new techniques to improve IoT safety and maximise computing resources. We develop a complete security architecture integrating lightweight cryptography, blockchain, machine learning anomaly detection, and federated learning. We did so because we know that traditional security measures are inadequate for the Internet of Things. The lightweight cryptographic algorithm (LCA) provides efficient encryption and decryption, making it ideal for low-resource Internet of Things devices. Twenty processes comprise the LCA design. These operations include key generation, data encryption, digital signatures, and integrity checking. These procedures secure IoT data transfers. ADML detects anomalies in encrypted Internet of Things data using machine learning. This approach may identify security issues better. To keep up with data trends, this method extracts features, trains models, and updates them. Blockchain-based data integrity (BDI) is the third element. Blockchain ensures that Internet of Things data is reliable and full. BDI developed an immutable ledger solution to increase IoT data security and dependability. This data integrity system generates blocks, hashes, confirms blocks, and updates the blockchain. Fourth, FLIoT (Federated Learning for the Internet of Things) emphasises data privacy and collaborative model training across IoT devices. Foundation for the Internet of Things (FIoT) protocols and standards aim to increase IoT devices' collective intelligence while safeguarding users' privacy. It includes local model training, model aggregation, and the latest global model distribution. Our work also uses Secure Multi-party Computation (SMC) to analyse data more thoroughly and continuously, addressing online transaction cybersecurity issues. The framework outperforms the current state of the art in memory use, energy consumption, anomaly detection accuracy and precision, and encryption and decryption time. The Hybrid Fusion Framework combines lightweight cryptographic algorithms with federated learning, machine learning, blockchain technology, and other similar technologies to provide an effective, adaptable, and affordable IoT security solution.