A Novel Homomorphic Encryption and Consortium Blockchain-based Hybrid Deep Learning Model for Industrial Internet of Medical Things
Aitizaz Ali,Muhammad Fermi Pasha,Antonio Guerrieri,Antonella Guzzo,Xiaobing Sun,Aamir Saeed,Amir Hussain,Giancarlo Fortino
DOI: https://doi.org/10.1109/tnse.2023.3285070
IF: 6.6
2023-01-01
IEEE Transactions on Network Science and Engineering
Abstract:Securing Electronic Medical Records (EMRs) is one of the most critical applications of cryptography over the Internet due to the value and importance of data contained in such EMRs. Although blockchain-based healthcare systems can provide security, privacy, and immutability to EMRs, several outstanding security and latency issues are associated with existing schemes. For example, some researchers have used the blockchain as a storage tool which increases delay and adversely affects the blockchain performance since it stores a copy of each transaction. A distributed ledger also requires appropriate space and computational power with increased data size. In addition, existing healthcare-based approaches usually rely on centralized servers connected to clouds, which are vulnerable to denial of service (DoS), distributed DoS (DDoS), and collusion attacks. This paper proposes a novel hybrid-deep learning-based homomorphic encryption (HE) model for the Industrial Internet of Medical Things (IIoMT) to cope with such challenges using a consortium blockchain. Integrating HE with the proposed IIoMT system is a vital contribution of this work. The use of HE while outsourcing to the cloud the storage provides a unique facility to perform any statistical and machine learning operation on the encrypted EMR data, hence providing resistance to collusion and phishing attacks. Our proposed model uses a pre-trained hybrid deep learning model in the cloud and deploys the trained model into blockchain-based edge devices in order to classify and train local models using EMRs. This is further conditioned on the private data of each edge and IoT device connected with the consortium blockchain. All local models obtained are aggregated to the cloud to update a global model, which is finally disseminated to the edge nodes. Our proposed approach provides more privacy and security than conventional models and can deliver high efficiency and low end-to-end latency for users. Comparative simulation analysis with state-of-the-art approaches is carried out using benchmark performance metrics, which show that our proposed model provides enhanced security, efficiency, and transparency.
engineering, multidisciplinary,mathematics, interdisciplinary applications