MUMBAI, India, Jan. 9 -- Intellectual Property India has published a patent application (202541113828 A) filed by Dr. Murugan K; Mahaswetha B; Keerthana S; Naveen M; Arikarthik S; and Dr. Pushpavalli M, Sathyamangalam, Tamil Nadu, on Nov. 19, 2025, for 'iot weather station using adapdive bayesian-lstm ensemble with contextual attention (able-ca) systems.'
Inventor(s) include Dr. Murugan K; Mahaswetha B; Keerthana S; Naveen M; Arikarthik S; and Dr. Pushpavalli M.
The application for the patent was published on Jan. 9, under issue no. 02/2026.
According to the abstract released by the Intellectual Property India: "The widespread development of Internet of Things (loT) technology has facilitated the development of smart weather monitoring systems with fine-grained, real-time, and reliable environmental information, Nevertheless, classical weather forecasting approaches are usually plagued by high uncertainty caused by noisy sensor readings, volatile climatic conditions, and a lack of contextual adaptability. To deal with these issues, this paper suggests an loT-powered Weather Station based on an Adaptive Bayesian-LSTM Ensemble with Contextual Attention (ABLE-CA) system. The system incorporates distributed loT-empowered weather stations with sensors for temperature, humidity, rain, wind speed, and atmospheric pressure, which send data continuously t,o a Cloud platform. The ABLE-CA modellev~rages the power of Bayesian irtference, Long Short-Term Memory (LSTM) networks, and attention mechanisms to produce highly accurate and uncertainty-aware predictions. The probabilistic estimation and prediction uncertainty quantification offered by the Bayesian framework are critical in weather-sensitive applications like agriculture, aviation, and disaster management. The ensemble method also improves robustness by combining multiple LSTM learners learned over different temporal parts of weather data. Meanwhile, the contextual attention mechanism adaptively allocates importance weights to pertinent environmental and temporal features, enhancing flexibility towards abrupt climate chan.ges. Experimental tests performed on actual meteorological data and loT-sensed data show that ABLE-CA dramatically outperforms traditional machine learning and isolated deep learning models in forecasting accuracy, robustness to missing values, and explainability. Furthermore, the system is optimized for real-time deployment with low-latency predictions and scalable integration over heterogeneous loT weather stations."
Disclaimer: Curated by HT Syndication.
