MUMBAI, India, July 24 -- Intellectual Property India has published a patent application (202611061562 A) filed by Sharda University on May 14, 2026, for Multi-Modal Federated Anomaly Detection System.
Inventors include Anand, Utkarsh; Singh, Raj Karan; Singh, Yashi; Sharma, Avinash Kumar; Nand, Parma; and Astya, Rani.
The application for the patent was published on July 17, 2026, under issue no. 29/2026.
Abstract: A multi-modal federated anomaly detection system (100) for Industrial IoT networks incorporates edge devices (102) equipped with multi-modal sensors (104) capture heterogeneous time-series data such as vibration, temperature, acoustic, and visual signals. This data is processed by a pre-processing and sync module (106), which handles noise filtering, time synchronization, and feature normalization. A local multi-modal anomaly detection module (108) fuses cross-modal data to generate anomaly scores. A self- adaptive threshold learning module (112) dynamically adjusts anomaly decision thresholds based on temporal behavioural drift, equipment aging, and environmental variability, producing anomaly decision signals that update the pre-processing and sync module (106). Additionally, a federated learning client (114), which encrypts model updates and transmits them to the federated aggregation server (110), ensures secure, distributed, and adaptive anomaly detection without sharing raw sensor data.
Disclaimer: Curated by HT Syndication.
