AI-driven security framework for medical sensor networks: enhancing privacy and trust in smart healthcare systems

AI-driven security framework for medical sensor networks: enhancing privacy and trust in smart healthcare systems

연구 분야: Artificial Intelligence

학회: Cluster Computing

Protecting medical sensor networks (MSNs) is important for the integrity and confidentiality of patient data. Traditional security methods often do not easily align with the dynamically changing cyber threat landscape and usually come with device resource constraints that are hard to meet by medical devices. This paper proposes an AI-driven intrusion detection system (IDS) using convolutional neural networks (CNNs) and lightweight authentication protocols to address these challenges. This study aims to design and assess a CNN-based IDS on MSNs, using lightweight authentication protocols to enhance real-time threat detection in secure communications. The research framework collected data from the NSL-KDD dataset, which consists of network traffic from various medical devices. Preprocessing was done to normalize the data and extract features from the dataset. The supervised-trained CNN model was tested through performance evaluation on several IDS evaluation metrics using accuracy, precision, recall, the area under the curve, and receiver operating characteristics (AUC-ROC). Lightweight authentication protocols were designed and analyzed for security and efficiency. As a result, the CNN-based IDS provides excellent NSL-KDD performance in terms of accuracy of 96.5%, precision of 95.9%, and recall of 96.7%, while the AUC-ROC is 0.982. The approach described shows a much greater accuracy and robustness than classical IDS solutions. Lightweight authentication protocols ensure safe communication with a low computational overhead, which is satisfactory for medical devices with limited resources. The proposed CNN-based IDS and lightweight authentication protocols significantly improve the security of MSNs. Therefore, they ensure that real-time and reliable threat detection and effective and secure communication are crucial requirements for data protection in healthcare.