Fog Computing Federated Learning System Framework for Smart Healthcare

Abstract

With the continuous growth of medical big data, the use of remote cloud servers for big data analysis becomes more and more complicated and difficult, and it is susceptible to security and privacy issues. In response to the above problems, a fog computing federated learning system for smart healthcare is proposed. The system uses the iFogSim simulation platform to establish a smart fog computing layer between sensor nodes and remote cloud servers to improve data analysis and processing capabilities; at the same time, the federated learning idea is introduced to integrate the federation. Combine learning and fog computing to build a fog federation framework to ensure the privacy of data interaction. Aiming at the redundancy and high dimensionality of medical big data, principal component analysis and variance analysis are used for preprocessing; fully considering the disaster tolerance of the system, an improved election algorithm and detection mechanism are proposed to improve the security of data interaction. Ensure the normal operation of the system. In this way, the problem of “data islands” and resource imbalance in the medical field can be solved. Using real cardiovascular data sets and simulated data sets for testing, the experimental results show that the fog computing federated learning system has a significant improvement in network usage, system delay, system energy consumption, and system accuracy.