Blockchain-Integrated Machine Learning Models for Secured IoT Data Sharing and Authentication

Abstract

The continuous growth of Internet of Things (IoT) devices has increased the demand for secure, scalable, efficient data sharing and authentication. Existing centralized architectures are increasingly exposed to cyber intrusions, data tampering, uncontrolled data breaches, and single points of failure. This research introduces BcML-IoT, an integrated framework for secure and intelligent IoT data exchange that incorporates blockchain technology and Machine Learning (ML) approaches to address these challenges. While blockchain actively serves as a decentralized, immutable log of activities with transparent transaction validation, ML can add an extra layer of security by providing better anomaly detection, device authentication, and applying secure and privileged access control through predictive analytics to identify potential rogue devices. Smart contracts can be used to automate the authentication process, and next-generation, lightweight consensus mechanisms help to reduce energy consumption and latency. The experimental evaluations show that BcML-IoT supports a high throughput of 85-87 Transactions Per Second (TPS), and unique block confirmation times under 4 seconds for up to 1,000 devices. When compared with traditional classifiers, ML-based anomaly detection can achieve 96.2% and 94.7% accuracy using XGBoost and LSTM respectively. BcML-IoT will detect all spoofing and replay attacks (100% detection), as well as 99.7% of data modification attacks with a FAR of 0.3%. The results show that BcML-IoT is low-latency, robust and secure for real-time IoT environments. It also provides strategies for developing decentralized, resilient and intelligent IoT ecosystems.