Software-Defined Networking (SDN) and Deep Learning (DL) for Safeguarding Health Data on Devices, including IoMT Devices

Abstract

The Internet of Medical Things (IoMT) has transformed modern healthcare by enabling remote patient monitoring, personalized treatment, and operational efficiency through interconnected medical devices such as smart insulin pumps and implantable cardiac devices. However, this rapid digitalization has significantly expanded the attack surface, exposing sensitive and immutable health data to escalating cybersecurity threats. Unlike financial data, compromised medical information and manipulated life-critical devices may result in irreversible consequences. Traditional security mechanisms, including perimeter-based firewalls and static encryption schemes, are increasingly inadequate in IoMT environments characterized by heterogeneous architectures, constrained computational resources, and long device lifecycles with limited patching capabilities. Additionally, the centralized reliance on cloud platforms introduces systemic risks, where a single compromised node can propagate network-wide disruption. This study proposes an SDN-enabled deep learning framework for proactive reconnaissance attack detection in IoMT networks. By leveraging Software-Defined Networking (SDN), network traffic is centrally monitored and dynamically controlled, while a Feed-Forward Neural Network (FFNN) analyzes real-time traffic features to identify scanning and probing activities at early stages. The model achieved a validation accuracy of 99.985% under experimental conditions. Furthermore, the framework was evaluated in a real-world simulated IoMT environment using Mininet, where the SDN controller facilitated both attack detection and mitigation. The results demonstrate the potential of integrating SDN with deep learning to create adaptive, scalable, and intelligent security architectures tailored for IoMT ecosystems.