MUMBAI, India, Aug. 12 -- Intellectual Property India has published a patent application (202631092344 A) filed by Abinash Panda; Naveen Kumar G N; Viji K; and Cmr Institute Of Technology Bengaluru on July 30, 2026, for Mesh-Based Localized Edge Inference System For Anomaly Detection In Iot Sensor Clusters.
Inventors include Abinash Panda; Naveen Kumar G N; and Viji K.
The application for the patent was published on August 07, 2026, under issue no. 32/2026.
Abstract: A mesh-based localized edge-inference system for anomaly detection in IoT sensor clusters is disclosed. A plurality of sensor nodes, each comprising sensors, a microcontroller, a lightweight on-device inference model, and a wireless mesh radio, perform local inference on acquired sensor data to generate anomaly flags. Rather than transmitting continuous raw sensor data, each node shares only the generated anomaly flag and minimal metadata over a low-bandwidth mesh network. A consensus module aggregates anomaly flags received from neighbouring nodes, applying optional weighting based on proximity, sensor type, or historical reliability, and confirms a system-level anomaly only when an aggregated score meets a consensus threshold, thereby reducing false positives. The system achieves low-latency, low-bandwidth, and privacy-preserving anomaly detection suitable for remote and industrial IoT deployments.
Disclaimer: Curated by HT Syndication.
