MUMBAI, India, Sept. 28 -- Intellectual Property India has published a patent application (202631100745 A) filed by Dr. Subhrapratim Nath; Mr. Aritrash Sarkar; Ms. Adrija Ghosh; Ms. Ankana Debnath; Mr. Pritam Mondal; Mr. Roheet Purkayastha; Dr. Sudipta Ghosh; and Meghnad Saha Institute Of Technology on August 20, 2026, for A Decentralized Multi-Modal Sensor Fusion Based Building Management System With Edge Intelligence And Hierarchical Control.
Inventors include Dr. Subhrapratim Nath; Mr. Aritrash Sarkar; Ms. Adrija Ghosh; Ms. Ankana Debnath; Mr. Pritam Mondal; Mr. Roheet Purkayastha; and Dr. Sudipta Ghosh.
The application for the patent was published on September 25, 2026, under issue no. 39/2026.
Abstract: The present disclosure relates to an intelligent decentralized building management and environmental automation system configured for real-time monitoring, occupancy-aware decision-making, adaptive environmental control, energy optimization, and security management using distributed sensing, edge-based artificial intelligence, and multi-modal sensor fusion. The disclosed system, referred to as the Active Decentralized Embedded Management System (ADEMS), comprises a plurality of embedded sensor nodes configured to acquire heterogeneous environmental and occupancy-related data including micromotion signals, thermal measurements, passive infrared sensing, luminosity data, gas concentration, temperature, humidity, and vision-based occupancy information. The system employs a hierarchical sensing and conditional computation framework wherein low-complexity sensing modalities continuously monitor environmental activity and selectively activate higher complexity inference stages based on contextual conditions, thereby reducing computational overhead and energy consumption while maintaining high detection reliability. Sensor data acquired from distributed nodes is transmitted to an edge-based control unit configured to perform preprocessing, probabilistic sensor fusion, contextual environmental analysis, occupancy estimation, and machine learning based inference. A lightweight convolutional neural network (CI.IN) deployed on an edge computing platform performs localized human detection and occupancy validation without dependence on cloud infrastructure, thereby reducing latency and improving privacy preservation. The system further implements adaptive decision-making algorithms configured to dynamically control lighting systems, HVAC systems, alarm mechanisms, and environmental acfuation modules based on computed contextual states and occupancy conditions. The disclosed architecture additionally supports modular deployment, scalable multi-node integration, distributed learning capability, and federated model aggregation mechanisms for collaborative system improvement across multiple sensor nodes. Experimental evaluation under realistic operating conditions demonstrates effective real-time operation, reliable occupancy detection, reduced false alarms through staged sensor validation, and robust probabilistic decision-making under noisy sensor environments. Accordingly, the present invention provides a scalable, energy-efficient, low-cost, and intelligent alternative to conventional centralized building management systems through the integration of distributed embedded sensing, adaptive contextual processing, edge artificial intelligence, and decentralized autonomous control mechanisms.
Disclaimer: Curated by HT Syndication.
