Blockchain-Based Privacy Preservation Scheme for Misbehavior Detection in Lightweight IoMT Devices

Area of Science:

• Cybersecurity in Healthcare
• Internet of Medical Things (IoMT)
• Blockchain Technology
• Federated Learning

Background:

• The Internet of Medical Things (IoMT) offers significant potential for improving healthcare delivery through remote diagnostics and enhanced user experiences.
• However, critical privacy and security vulnerabilities persist in IoMT systems, hindering widespread adoption.
• Existing misbehavior detection methods often lack robust privacy preservation and are susceptible to contextual manipulation, leading to false positives.

Purpose of the Study:

• To develop an efficient and privacy-preserving framework for secure misbehavior detection in lightweight IoMT devices.
• To address the limitations of current methods by integrating blockchain and federated learning for enhanced security and privacy.
• To specifically apply and evaluate the framework within the context of an artificial pancreas system (APS).

Main Methods:

• Implementation of a privacy-preserving bidirectional long-short term memory (BiLSTM) model for misbehavior analysis.
• Integration of blockchain technology, specifically an Ethereum smart contract environment, to augment system security.
• Empirical benchmarking of the framework focusing on privacy preservation, incentive mechanisms, untraceability, and detection performance.

Main Results:

• The proposed framework demonstrates a high recall rate of 99.93%, indicating effective detection of nearly all malicious events.
• The system ensures sustainable privacy preservation and incorporates an untraceable incentive scheme.
• Analysis of resource consumption shows an average gas consumption of 84,456.5 and an Ether cost of 0.03157625.

Conclusions:

• The combined use of blockchain and federated learning offers a powerful solution for secure and privacy-preserving misbehavior detection in IoMT.
• The proposed framework significantly enhances the security of critical healthcare systems like the APS, minimizing risks associated with malicious activities.
• The model's high detection rate and privacy features make it a promising approach for securing the future of connected healthcare.