MUMBAI, India, Aug. 10 -- Intellectual Property India has published a patent application (202611074606 A) filed by Praduman Amroliya; and Santosh Kumar Sharma on June 16, 2026, for Hybrid Deep Reinforcement Learning Control For Efficient Wideband Mimo-Wpt Via Dynamic Wave-Front Control.
Inventors include Praduman Amroliya; and Santosh Kumar Sharma.
The application for the patent was published on July 31, 2026, under issue no. 31/2026.
Abstract: The present invention relates to a frequency-agile wideband wireless power transfer system incorporating multi-input multi-output transmission, programmable metasurface wave-front control, and hierarchical deep reinforcement learning for adaptive energy delivery. The system comprises a wideband radio frequency source, a multi-element transmitter array, a programmable transmissive metasurface, a rectenna-based receiver subsystem, and a hybrid controller configured to jointly optimize operating frequency, transmitter excitation phases, and metasurface phase states. A Deep Q-Network performs discrete frequency-band selection while a Proximal Policy Optimization module executes continuous phase optimization. The controller dynamically adapts to environmental variations, receiver movement, channel disturbances, blockage events, and frequency-dependent coupling conditions. The architecture utilizes electromagnetic response data, scattering parameters, and feedback signals to maintain power transfer efficiency across a wide operating spectrum. The invention further enables adaptive beam focusing, dynamic frequency hopping, mutual coupling suppression, and spatial power uniformity for wireless power distribution applications.
Disclaimer: Curated by HT Syndication.
