Blockchain-based edge computing for medical data storage & processing using federated learning

Abstract

With a great number of IoT devices being used in healthcare and a massive rise in medical data produced by these devices, data storage and processing systems using the traditional cloud computing framework are not enough to meet real-time data re- requirements in Internet-based services as data is transferred to faraway cloud servers for processing, resulting in high latency and costs. Edge computing can provide a solution to this problem by effectively offloading a portion of the workload from the cloud to nearby edge servers to perform data processing tasks close to the end-users, thus reducing latency and cost as well as improving the quality of service. However, edge computing faces threats regarding data privacy and security due to edge nodes being more vulnerable to cyber-attacks. To address this problem, blockchain can be integrated to protect data from tampering, maintain data integrity, and allow reliable access, distributed computation, and decentralized data storage. Thus, in this research, we present a secure medical data storage and processing system using blockchain and edge computing to preserve our clients’ data privacy. To tackle privacy and security concerns, federated learning using a neural network has been used to train models locally using the data on the edge nodes rather than sending relevant private information to a centralized server for training, and model parameters, as well as IPFS file hashes and other private information, are securely stored on the blockchain by incorporating cryptographic techniques.