Deep learning for cryptanalysis attack on IoMT wireless communications via smart eavesdropping
Zakaria Tolba, Makhlouf Derdour · 2021
The Internet of Medical Things (IoMT) faces grave protection concerns due to weakness related to the leakage of sensitive data controlled by detectors and transferred to the cloud through open wireless connections.In most IoMTs, patients information is encrypted by terminal devices, and the published cryptanalysis works demonstrate apparent limitations because they are performed for the theoretical model without any real implementation scenario on wireless encryption protocols, which poses a performance gap between the hypothetical investigation and practical employment of these attacks. To overcome these drawbacks, we suggest an approach based on a deep learning model for an attack simulation on specific IoMT protocols that discovers the private key of IoMT wireless communications using a smart eavesdropping conceptual scenario on multilevel layers of IoMT protocol stack for its implementation. We detail the attack process for building the datasets from wireless packet protocols with the aim of training the proposed cryptanalysis model.