MUMBAI, India, Aug. 10 -- Intellectual Property India has published a patent application (202641090994 A) filed by Vellore Institute Of Technology on July 27, 2026, for Autonomous Iot Mesh Topology Control Using Energy-Harvesting-Aware Reinforcement Learning.
Inventors include Ragavan K; and P N Bhargav Teja.
The application for the patent was published on July 31, 2026, under issue no. 31/2026.
Abstract: ABSTRACT AUTONOMOUS IOT MESH TOPOLOGY CONTROL USING ENERGY-HARVESTING-AWARE REINFORCEMENT LEARNING A method for autonomous topology control in a wireless mesh network comprising a plurality of IoT nodes (100), each comprising a wireless transceiver (102), a microcontroller (104), a non-volatile memory (106), and an energy harvesting circuit (108) coupled to a power monitor (110). The microcontroller (104) reads harvest power data and a state-of-charge value from an energy storage unit (112), computes a harvest budget as a surplus value, computes a discharge rate over a sliding window of beacon epochs, and predicts a time-to-empty value. Responsive to the time-to-empty value falling below a warning threshold, a handoff request packet is broadcast to one-hop neighbour nodes. The microcontroller (104) performs a lookup in a compressed Q-table (114) having a bounded size, receives a gossip packet (116) encoding a best action identifier and Q-value, blends local Q-values using a weighted update rule, selects an action, and writes to a transmit power control register or a role flag in the non-volatile memory (106).
Disclaimer: Curated by HT Syndication.
