A lightweight multi-vector DDoS detection framework for IoT-enabled mobile health informatics systems using deep learning

Abstract

The exponential growth in adopting Internet of Things (IoT) applications and services has rendered IoT security an essential concern that must be handled promptly. Multi-vector Distributed Denial of Service (DDoS) attacks are more intensified forms of DDoS attacks, and the anomaly-based Intrusion Detection System (IDS) schemes are the best suitable for detecting and mitigating them. However, deploying anomaly-based IDS frameworks in healthcare systems is particularly difficult since it involves longer processing times, increased complexity, and the need to preserve temporal features. This study presents a novel anomaly-based IDS framework that utilizes proposed stacked modified Gated Recurrent Units (mGRU) to detect and identify the Multi-vector DDoS attacks in mobile healthcare informatics systems. In order to generate user-specific results, we have developed two instances of IDS, namely the Binary Classification Engine (BCE) and the Multi-label Classification Engine (MCE). The empirical results demonstrate that the proposed mGRU-based IDS models outperform the standard GRU-based IDS models, with a reduction in time consumption of around 2% on the CICIoT2023 and CICDDoS2019 datasets. The proposed IDS instances provide leading-edge metrics, lightweight features, and user-specific results, making them suitable for effective deployment in time-critical healthcare applications and services.