PEF-CAPD: A Privacy Enhanced Federated Cyber Physical and Attack Detection Framework for Edge-Cloud-Blockchain Enabled Smart Healthcare Environment

Abstract

Recently, healthcare industries faced severe cybersecurity problems due to the widespread amalgamation of technologies into a smart healthcare environment. As the number of attacks increased, the crucial healthcare sectors were targeted by cyber attackers. Conventional cybersecurity operations were not very effective due to their heterogeneity and complexity, respectively. In this research, we propose a novel privacy-preserving and attack detection framework named Privacy Enhanced Federated Cyber Physical and Attack Detection (PEF-CAPD) for the healthcare environment. The proposed research exploits edge computing, cloud computing, and federated learning technologies, respectively, to enhance the applicability and privacy in the healthcare environment. Initially, the medical data from the medical devices are securely encrypted and provided to the CMS. Note that the medical devices are connected in a MESH structure to enable self-healing, scalability, and reliability properties. In the CMS, the collected data are subjected to pre-processing, in which the pre-processed data are fed to the ES, where the patient-specific local models are generated using Skipped Dense Neural Network (SDNN) from local attack detection datasets. The generated local models are provided to the BCS for global model aggregation using the Novel Federated Aggregation Model (NFAM). From the aggregated global model, the Advanced Explainable Support Vector Machine (AEX-SVM) detects the possible attacks in the healthcare environment. The proposed work is validated on benchmark datasets that are generated from varied healthcare environments. The validation results show that the proposed approach demonstrates noteworthy accuracy of 99.92% compared to the state-of-the-art works.