Privacy Preservation In Iot Networks Through Dehaene–Changeux Model Integrated With Graph Convolutional Layers And Non-Linear Analysis

Abstract

The fast development of the Internet of Things (IoT) has brought privacy preservation in distributed networks more and more of a problem. Sometimes the complex, dynamic interactions defining these systems are not handled by conventional methods of IoT network security. Combining Graph Convolutional Layers (GCL) with Non-Linear Analysis (NLA) and the Dehaene–Changeux Model (DCM), this work offers a novel approach to circumvent these restrictions and hence improve privacy protection in IoT systems. Originally employed to mimic and assess the information flow inside IoT networks, the DCM is changed in this work to show cognitive processes. While GCL assists the model to effectively collect and evaluate the complex network topologies common of IoT contexts, NLA is utilized to identify and lower non-linear dangers to data privacy. The synthetic IoT dataset employed in evaluating the proposed approach consisted in ten thousand nodes and fifty thousand edges, therefore simulating several real-world attack scenarios. Privacy violations were found with a 92.3% accuracy using GCL paired with DCM, 15.4% more accurate than more traditional methods. Moreover, NLA greatly reduced the false positive rate to 3.7%, thereby enhancing the dependability of the privacy-preserving system. The results reveal that combined DCM, GCL, and NLA not only raises IoT network resilience of privacy preservation but also offers a scalable real-time application solution. This work prepares the ground for additional research on cognitive-inspired models for improved security solutions in IoT environments.