MUMBAI, India, March 13 -- Intellectual Property India has published a patent application (202641024381 A) filed by Vellore Institute Of Technology, Vellore, Tamil Nadu, on March 1, for 'multi-modal edge-based system and a method for real-time seismic anomaly detection on edge devices.'

Inventor(s) include Valarmathi B; Srinivasa Gupta N; Ansh Jerath; and Sanuj Sudhakar Dhote.

The application for the patent was published on March 13, under issue no. 11/2026.

According to the abstract released by the Intellectual Property India: "The present invention relates to a multi-modal edge-based seismic anomaly 5 detection system and method are disclosed. the system comprises a primary edge device including a processor and a microphone configured to acquire ambient acoustic pressure signals, and a secondary edge device including an inertial measurement unit (IMU) configured to acquire kinematic motion data. a low- latency local network interface enables transmission of the kinematic motion data 10 from the secondary edge device to the primary edge device. a sensor fusion engine executed on the primary edge device converts the acoustic pressure signals into time-frequency representations and applies a trained convolutional neural network to identify seismic P-wave acoustic signatures, while independently evaluating kinematic motion energy derived from the IMU data. a seismic alert is generated 15 only when both the acoustic signature and the kinematic motion energy exceed predefined thresholds, thereby suppressing false positives caused by non-seismic acoustic or motion events. the system employs Boolean AND-based fusion logic to reject domestic noise sources in the absence of corresponding motion and to reject kinematic disturbances lacking seismic acoustic characteristics. a 20 corresponding edge-based method includes max-energy pooling down-sampling of acoustic signals, spectrogram generation, convolutional neural network classification, kinematic energy computation, and local validation of seismic events without reliance on cloud infrastructure. validated events may be recorded as time-stamped acoustic and motion data for post-event analysis."

Disclaimer: Curated by HT Syndication.