MUMBAI, India, July 7 -- Intellectual Property India has published a patent application (202641063684 A) filed by Nandha Engineering College on May 20, 2026, for Ai Powered Cyber Attack Detection In Industrial Iot Using Dbn And Cnn Integration.
Inventors include Ms. K Gobika; Ms. A Hemapriya; and Ms N Zahira Jahan.
The application for the patent was published on July 03, 2026, under issue no. 27/2026.
Abstract: An intelligent cybersecurity system and method for detecting and preventing cyber attacks in Industrial Internet of Things (lloT) environments is disclosed, employing a hybrid deep learning architecture that integrates a Deep Belief Network (DBN) with a Convolutional Neural Network (CNN). Industrial loT networks deployed across smart manufacturing and automated industrial systems are inherently vulnerable to cyber threats including Denial-of-Service (DoS), network spoofing, and data tampering, against which conventional intmsion detection systems demonstrate inadequate detection capability due to the complexity and scale of IloT data streams. The proposed system continuously ingests real-time network traffic, sensor telemetry, and device conununication logs fi·om interconnected IloT nodes and subjects the colJected data to an intelligent preprocessing pipeline comprising normalization, statistical feature extraction, and temporal sequence encoding. The DBN component employs stacked Restricted Boltzmann Machines to perform unsupervised hierarchical feature learning, extracting deep latent representations that capture the statistical baseline of normal industrial network behaviour. The extracted feature representations are subsequently forwarded to the CNN component, which applies convolutional filtering and max-pooling operations to identify spatial attack patterns and performs multi-class classification distinguishing normal traffic from malicious behaviors including DoS floods, spoofing attempts, and data integrity violations. The hybrid DBNCNN architecture synergistically combines the generative anomaly detection strength of the DBN with the discriminative pattern recognition capability of the CNN, achieving superior detection accuracy, enhanced precision and recall, and significantly reduced false alarm rates compared to conventional machine learning approaches. Upon confinned threat detection, the system automatically generates risk-scored alerts, executes adaptive firewalJ policy updates, and issues real-time security recommendations to industrial administrators. Experimental evaluation demonstrates that the proposed DBN-CNN model outperforn1s existing intrusion detection methods across ali standard classification metrics, providing a reliable, scalable, and adaptive cybersecurity solution for securing critical Industrial loT infrastructure.
Disclaimer: Curated by HT Syndication.
