Deep Reinforcement Learning-Based Cyber Defense Mechanism for Intelligent Threat Mitigation in Healthcare IoT Networks

Abstract

Healthcare Internet of Things (HIoT) network has greatly enhanced real-time patient monitoring, remote diagnostic, and smart healthcare management, but has created more weaknesses to advanced cyberattacks including distributed denial-of-service (DDoS) attacks, malware injection attacks, unauthorized access, and manipulation of data attacks. Conventional intrusion detection systems usually use any of the following: static subscription and rule-based methods, which prove insufficient to deal with dynamic and changing cyber threats in heterogeneous healthcare IoT settings. To undo these drawbacks, the proposed study will present a Deep Reinforcement Learning-based cyber defense system with a Deep Q-Network (DQN) system to detect threats and mitigate them intelligently in healthcare IoT networks. The proposed framework constantly acquires the best defense mechanisms by engaging with network settings and dynamically chooses the right mitigation measures towards malicious behavior. The TON IoT and CICIDS2017 datasets with varied normal and attack traffic scenarios were used to run the experimental evaluation. Classification metrics such as Accuracy, Precision, Recall, F1-Score, and False Positive Rate (FPR) were used to perform performance analysis. The experimental outcomes prove that the suggested DQN-based system shows better intrusion detection and threat management capabilities, and the lower occurrence of false alarms than conventional machine learning-based intrusion detection methods. The suggested intelligent cyber defense architecture is a step towards the creation of safe, resilient, and scalable healthcare IoT ecosystems, since it incorporates deep reinforcement learning to manage real-time autonomous cybersecurity.