Enhancing IIoT Security using KNN based Hypergraph clustering through Zero Trust Micro-Segmentation for Dynamic network protection

The proliferation of IoT devices has led to a dramatic increase in security breaches, highlighting the need for advanced security measures. This paper investigates the implementation of zero-trust architecture (ZTA) with a focus on micro-segmentation using machine learning algorithms. Microsegmentation involves dividing the network into groups based on similar traffic patterns, thus limiting network traffic and hindering lateral movement by intruders. This granular approach reduces unauthorized access and data breaches are minimized by ensuring that each segment has independent security. The proposed method uses deep non-symmetric autoencoder for feature extraction, generates kNearest Neighbors (k-NN) based hypergraphs to capture complex data relationships, and clustering algorithms such as DBSCAN, Incremental K-Means, OPTICS, ART, and CluStream are used to identify and isolate microsegments. By leveraging knn based hypergraph for cluster formation, reducing the lateral movements and enhancement of IoT security is achieved.