Intellectual Property India Publishes Patent Application for Hybrid Machine Learning Based Doppler-Aware Channel Estimation System For Cooperative Underwater Acoustic Networks

MUMBAI, India, Aug. 10 -- Intellectual Property India has published a patent application (202621073872 A) filed by Symbiosis International Deemed University on June 12, 2026, for Hybrid Machine Learning Based Doppler-Aware Channel Estimation System For Cooperative Underwater Acoustic Networks.

Inventors include Ranvir Thakur; Tanmay Lende; Aripaka Balaji; and Rajeshwar L Balla.

The application for the patent was published on July 31, 2026, under issue no. 31/2026.

Abstract: ABSTRACT HYBRID MACHINE LEARNING BASED DOPPLER-AWARE CHANNEL ESTIMATION SYSTEM FOR COOPERATIVE UNDERWATER ACOUSTIC NETWORKS The present invention discloses a hybrid machine learning based Doppler-aware channel estimation system (100) for cooperative underwater acoustic networks. The system (100) comprises a Doppler-aware channel modelling module (110) that generates realistic underwater acoustic signal realizations incorporating Doppler frequency shifts, multipath propagation, and additive noise. A data preprocessing module (120) applies sliding window based feature extraction. A machine learning estimation engine (130) comprising Sparse Bayesian Learning (131), LSTM (132), Random Forest (133), Multilayer Perceptron (134), and hybrid SBL-LSTM (135) estimators reconstructs clean waveforms from distorted observations. A cooperative multi-node architecture module (140) enables configurable topologies from single-node through ten-node configurations exploiting spatial diversity. A performance evaluation engine (150) quantifies accuracy using RMSE, NMSE, BER, Correlation, and SNRi metrics. The SBL estimator achieves the lowest RMSE of 0.1117 and highest correlation of 0.9648. [

Disclaimer: Curated by HT Syndication.