MUMBAI, India, Sept. 22 -- Intellectual Property India has published a patent application (202641108557 A) filed by Easwari Engineering College; and Srm Institute Of Science And Technology,ramapuram Campus on September 10, 2026, for A Federated Learning-Based System For Detecting Intrusions In Iot Networks Using Cnn.
Inventors include Nagha Narasimha G; Kiran Teja V; and Saranya D.
The application for the patent was published on September 18, 2026, under issue no. 38/2026.
Abstract: The present invention is an artificial intelligence-based system for real-time, privacy-preserving intrusion detection across distributed Internet of Things (IoT) networks. The system integrates federated learning, CNN-based classification, bio-inspired feature selection, and robust model aggregation. It preprocesses MQTT-based IoT network traffic, selects an optimal feature subset and trains lightweight CNN models locally at each IoT gateway without transmitting raw data to a central server. Local model updates are screened by a malicious client filtering mechanism that excludes updates deviating significantly from the median, then aggregated using Federated Averaging with Momentum to iteratively improve a global detection model. Network traffic is classified into legitimate traffic and multiple distinct attack categories, including Brute Force, Denial of Service, Flood, Malformed data attacks, enabling identification of the specific intrusion type. The invention offers a scalable, computationally efficient, and adversarially resilient solution for securing heterogeneous IoT networks. Total Number of Words: 141 words
Disclaimer: Curated by HT Syndication.
