Deep Reinforcement Learning Aided Device Profiling for Anomaly Detection in IIoT Over Zero Trust Security Network

The application of internet of things in industries is the fourth revolution of industries and is referred to as industry 4.0 or industrial internet of things. The shifting of manual and conventional approach to a digital platform is witnessed as a key consideration for the emergence of industrial internet of things. Industrial internet of things works around thousands of devices connected remotely and the overall monitoring, functioning and storage is performed virtually by and on devices, hence more prone to cyber threads. With the amplification in the growth and application of more complicated networks, the requirements for the security are also increased significantly. Zero trust security stands as an efficient solution for all the security questions. Zero trust security involves continuous monitoring of devices for verifying the authenticity before permitting the access to network. Continuous monitoring for detecting the honest and malicious device adversely effects the productivity of the IIoT environment. The proposed paper presents a novel approach to generate a unique profile of devices via deep reinforcement learning to avoid irrelevant authentication cycles and produce more optimized results. The proposed work surpasses the existing approaches (CRH-SRP) [1] and Cyber Security via Determinism [2] with enhanced data confidentiality rate by 3.25% and reduced false positive rate by 30.5% respectively.