A Self-Adaptive Framework for Responding to Uncertainty in Access Control Process with Deep Neural Networks

The rapid increase in Internet of Things (IoT) devices and data across various domains (e.g., healthcare, smart cities, smart factories, and smart homes) has led to growing complexity in managing security policies to protect them. Moreover, the increased complexity makes it even more challenging to define security policies in the attribute-based access control (ABAC) model, which requires considering user requirements and environmental conditions. Thus, the ABAC model is likely to include unexpected uncertainties when administrators define security policies. In this study, uncertainty refers to unforeseen future outcomes that may arise from factors the administrator has not considered. Furthermore, the increasing number of IoT devices makes it nearly impossible for administrators to predict all uncertainties and adjust security policies accordingly. Several studies have been proposed to address these challenges, but they rely on human intervention. Therefore, this study proposes a self-adaptive framework that combines a self-adaptive system with deep neural networks (DNNs) to respond to uncertainties intelligently and autonomously. The framework effectively identifies uncertainties in the access control process through variational autoencoder (VAE) and long short-term memory (LSTM) without human intervention and autonomously responds using a self-adaptive system.