Hierarchical Federated Learning-Based Method for privacy-preserving in the Healthcare Environment
Manoj Kumar D P, Ananda Babu Jayachandra, Jaladi NagaSireesha, Bhattu Venkatesh, Ayesha Siddiqua · 2023
The COVID-19 pandemic has led to increase the demand for medical equipment with internet of things (IoMT) capabilities forcing doctors to diagnose and treat patients remotely. This has increased awareness of health, leading to a surge in purchasing of IoMT-capable medical equipment. However, medical data is sensitive and valuable on the dark web, making it vulnerable to cybercriminals. Due to limited storage and processing capabilities of network devices, system administrators struggle to enhance security measures. This work aims to investigate rapidly spreading attacks before they compromise wellness systems. Hierarchical federated learning is (HFL) proposed using a model built on Dew-cloud, which offers improved data privacy due to the increased availability of IoMT critical applications. Dew servers with distributed architecture and a cloud computing-backed backend implement hierarchical long-term memory (HLSTM) method. The implemented HFL-HLSTM method’s performance is evaluated using NSL-KDD dataset. Implemented method achieved better performance than the existing intrusion detection methods including MiTed, LS2DNN, and MetaCID and the metrics are 98.62% of f-measure, 99.53% of accuracy, 99.21% of precision, and 99.10% of recall.