Protecting Healthcare Supply Chain Infrastructure from DOS Attacks with Anomaly Detection

Abstract

The digital universe has increased the risk of complex DoS attacks against supply chain systems; consequently, there is a substantial chance of a healthcare infrastructure being compromised by unavailability due to a DoS attack. In this paper, we provide a detailed machine learning-based method for detecting and mitigating those complex DoS attacks in the healthcare supply chain infrastructure ~ via the incorporation of four separate machine learning-based models: Random Forest, XG-Boost, Isolation Forest, and Deep Autoencoder to identify and classify network services and differentiate and detect multiple types of DoS attacks, e.g., SYN Flood, UDP Flood, HTTP Flood, ICMP Flood, etc. The proposed system has been shown to produce results with very high accuracy (>99.5%), precision (99.10 – 99.50), and detection time (<5 ms). In addition, the proposed system has prevented the OrdersCleaned.csv dataset from being accessed by unauthorized users, thereby enabling legitimate users to use the dataset without interruption. The results from the proposed system make a contribution to the area of healthcare supply chain cybersecurity and provide an effective, efficient, and scalable means of providing real-time protection for the healthcare supply chain infrastructure.