MUMBAI, India, Jan. 2 -- Intellectual Property India has published a patent application (202541123147 A) filed by Malla Reddy (MR) Deemed to be University; Malla Reddy College Of Engineering And Technology; Malla Reddy Engineering College For Women; Malla Reddy University; and Malla Reddy Vishwavidyapeeth, Medchal-Malkajgiri, Telangana, on Dec. 6, 2025, for 'federated ai system for smart farming decision support.'
Inventor(s) include Dr. Shaik Javed Parvez; G. Gayatri; Mr. Jakkula Prema Sagar; Dr. R. Sivasubramanian; and Ms. Noureen Tabassum.
The application for the patent was published on Jan. 2, under issue no. 01/2026.
According to the abstract released by the Intellectual Property India: "The current invention reveals a new Federated Artificial Intelligence (AI) System that is specially designed to deliver effective decision support to precision agriculture without violating the privacy of the data of individual farming organizations too harshly. The unwillingness of farmers to offer proprietary data (soil health metrics, yield history, pest infestation rates, etc.) to centralized cloud platforms is a common barrier to the adoption of data-driven technologies in the modern agricultural landscape because of the concern regarding the commercial sensitivity of data and its data sovereignty. Traditional smart agriculture applications are based on sending raw data to central servers to be processed, which in addition to a privacy risk poses a high bandwidth requirement that is often unavailable in rural and remote agricultural areas. To address these severe restrictions, the invention adopts the concept of a decentralized machine learning architecture that employs the Federated Learning (FL) concepts. The proposed system allows the local machine learning models to be trained directly on edge computing devices at the farm level, i.e. smart tractors, drone base stations, or localized IoT gateways, instead of transmitting raw sensor data to a central cloud. This set of local nodes computes an update to the model, namely, gradient information or change of the synaptic weights, but encrypts them and transfers them to a central aggregation server. The server synthesizes such updates to enhance an overall crop management model without needing to see the underlying raw data which in turn creates a collective intelligence which is beneficial to all concerned with a high level of data isolation. Also, the system is inclusive of a localized Decision Support System (DSS) which uses the globally aggregated model to give real-time and context-sensitive suggestions to the farmer. This DSS can be used in offline as well as online modes, although the locally stored model analyzes the inputs of soil sensors, weather stations, and satellite images. The edge processing of data greatly diminishes latency enabling the system to execute instant actuation or precision spraying machinery of irrigation systems or precision spraying equipment depending on the changing environmental conditions. The architecture also takes into consideration the issue of heterogeneity of data in the agricultural sector where various farms can have different types of sensors and collection formats. In the invention, an adaptive normalization layer is used to normalize the inputs in the local level during the training process to ensure that the global model is resilient to different farming ecosystems. This democratic method makes high-tech AI agricultural data access more accessible to small-holder farms, and such high-tech features can eventually be used to predict upcoming yields and maintain resource sustainability by larger industrial farming operations."
Disclaimer: Curated by HT Syndication.
