MUMBAI, India, Aug. 31 -- Intellectual Property India has published a patent application (202641099829 A) filed by Mr. Y. Jeevan; Dr. V. Saidulu; Gudipati Varshitha; Mrs. Avirineni Phanimadhuri; Naga Venkateshwara Rao K; Dr. Srinivasan Nagaraj; Mrs. P. Jessie; and Mrs. M. Abinaya on August 18, 2026, for An Ai-Assisted Low-Power Iot Communication Framework With Adaptive Spectrum Utilization.

Inventors include Mr. Y. Jeevan; Dr. V. Saidulu; Gudipati Varshitha; Mrs. Avirineni Phanimadhuri; Naga Venkateshwara Rao K; Dr. Srinivasan Nagaraj; Mrs. P. Jessie; and Mrs. M. Abinaya.

The application for the patent was published on August 28, 2026, under issue no. 35/2026.

Abstract: ABSTRACT OF THE INVENTION An artificial-intelligence-assisted low-power Internet of Things communication framework with adaptive spectrum utilization is disclosed. The framework integrates energy-efficient spectrum sensing, neural-network-based occupancy prediction, and multi-agent reinforcement learning to dynamically select communication channels, transmit power levels, and modulation schemes for battery-constrained IoT devices. Spectrum sensing is performed at a low duty cycle using compressed sensing techniques followed by a compact convolutional neural network, while a recurrent neural network forecasts future occupancy. Each IoT node acts as an independent agent in a multi-agent deep Q-network that optimizes a composite reward balancing packet delivery success, energy efficiency, and fairness. The learned policy is quantized and deployed on resource-limited microcontrollers, enabling real-time decisions with minimal memory and computational overhead. A complementary adaptive duty-cycling medium- access control protocol further reduces idle listening and control-packet energy expenditure. Experimental results on a large-scale test- bed demonstrate approximately 47 percent reduction in energy consumption and packet delivery ratios exceeding 98 percent under realistic interference conditions. The framework supports heterogeneous radio technologies, hierarchical clustering for scalability, and offline model updates, making it suitable for dense smart-city, industrial, and agricultural deployments where long battery life and reliable connectivity are essential. The invention therefore provides a practical, standards-compatible solution for sustainable large- scale IoT networks.

Disclaimer: Curated by HT Syndication.