Improving Access Control Decisions using Deep Learning and Contextual IoT Features

Abstract

Securing IoT environments has become more complex because it requires the combination of several technologies, from physical devices and wireless transmission to mobile and cloud architectures. Traditional access control models that rely solely on credentials or roles are inflexible and ineffective in intelligent and dynamic environments. To address these limitations, we proposed a more advanced access control model that exploits the power of machine learning to solve problems relating to access control decision-making. In particular, we propose Deep Learning context Based Access Control (DLCBAC) byleveraging significant advances in deep learning technology as a potential solution to this problem. The context-aware approache enable fine-grained control over data and resources, tailoring access permissions based on the specific circumstances surrounding the access request.We experiment with a real-world dataset collected from real IoT envirenoment. Our evaluation results suggest that a DLCBAC could recommend granting or denying permission with 99.9% accuracy.