MUMBAI, India, July 24 -- Intellectual Property India has published a patent application (202611057882 A) filed by Panipat Institute Of Engineering And Technology Piet on May 06, 2026, for System And Method For Predictive Edge-Cloud Resource Scaling In Federated Learning Environments.
Inventors include Dr. Sakshi Patni; Mr. Anurag Vashist; Ms. Rajni; and Dr. Dinesh C. Verma.
The application for the patent was published on July 17, 2026, under issue no. 29/2026.
Abstract: The present invention a system and method are disclosed for intelligent edge-cloud resource scaling in federated learning environments. Edge nodes execute local training and inference within isolated execution instances and continuously collect non- sensitive telemetry including CPU utilization, memory usage, queue length, task arrival rate, inference latency, and training duration. A workload manager classifies incoming tasks and assigns priority levels. A demand prediction engine forecasts near-future compute demand using telemetry sequences and, in a preferred embodiment, employs a spatio-temporal learning model. A scaling decision module proactively applies vertical scaling, horizontal scaling, and selective cloud offloading for permitted tasks to maintain latency and stability. A straggler mitigation module reduces federated training round delays by predicting slow participants and allocating additional resources or replicas to align completion times. A federated learning coordinator performs secure aggregation of model updates while preserving privacy by keeping raw datasets at the edge. Audit logs record scaling decisions and allocations for governance. (Accompanied Figure No. 1)
Disclaimer: Curated by HT Syndication.
