Doppler Spread Estimation Using Supervised Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current Doppler spread estimation methods in wireless communications, such as those using inverse Bessel functions, are inaccurate due to assumptions about channel statistics and sensitivity to noise variance, leading to low resolution and inaccuracy, especially in varying signal-to-noise ratios and millimeter wave frequencies.
Innovation Solution
Employing supervised machine learning, specifically multi-layer perceptron neural networks trained on diverse data sets, to predict Doppler spread from estimated channel correlations, allowing for accurate estimation across different operating conditions and noise levels by selecting appropriate predictors based on signal-to-noise ratio and Doppler shift ranges.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If inverse Bessel function methods are used for Doppler spread estimation, then the estimation can be performed with simple mathematical operations, but the accuracy and resolution are low due to assumptions about channel statistics and sensitivity to noise variance
Solution Approach 1:
The patent replaces traditional mathematical inversion methods (inverse Bessel functions) with a machine learning-based neural network system. The neural network is trained offline to learn the complex mapping between channel correlation and Doppler spread, substituting the mathematical inversion process with a trained predictor that provides higher accuracy without requiring explicit mathematical formulas during operation.
Solution Approach 2:
The patent performs preliminary training of the neural network predictor offline using simulated channel data across various Doppler spreads and signal-to-noise ratios. This preliminary action creates a pre-trained model that can be deployed in the receiver, allowing accurate Doppler spread estimation without real-time computational complexity. The training phase prepares the system to handle varying channel conditions efficiently during actual operation.
2Adaptability or versatility
If traditional channel estimation methods are used, then the system operates with fixed algorithms, but the adaptation to varying channel conditions and noise levels is insufficient
Solution Approach 1:
The patent implements a dynamic adaptation mechanism where the system selects different pre-trained neural network predictors based on the current signal-to-noise ratio conditions. The receiver estimates the SNR and chooses the appropriate predictor from a set of predictors trained on different SNR ranges, allowing the system to adapt to varying channel conditions and maintain high reliability across different operating environments.
Solution Approach 2:
The patent changes the operational parameters of the estimation system by using multiple neural network predictors trained on different signal-to-noise ratio ranges and Doppler shift ranges. Instead of using a single fixed algorithm, the system dynamically adjusts which predictor is used based on current channel parameters, thereby improving adaptability to varying conditions while maintaining communication reliability.
3Measurement precision
If a single neural network predictor is used across all conditions, then the device complexity is reduced, but the accuracy varies significantly across different signal-to-noise ratios and Doppler shift ranges
Solution Approach 1:
The patent segments the operating range into different signal-to-noise ratio ranges and Doppler shift ranges, with each segment having a dedicated pre-trained neural network predictor. This segmentation allows each predictor to be optimized for its specific range, maintaining high accuracy across all conditions. The receiver divides the overall operational space into manageable segments and selects the appropriate predictor based on current conditions.
Solution Approach 2:
The patent applies local quality by training different neural network predictors with different characteristics for different operating conditions. Each predictor is locally optimized for its specific SNR and Doppler shift range, rather than using a single general-purpose predictor. This local optimization ensures high estimation accuracy for each specific condition while maintaining overall system performance across all conditions.
Data Source
AI summary
A radio receiver includes a channel estimator processing circuit including: a feature extractor configured to extract one or more features from a received signal, the features including a channel correlation estimated based on a reference signal in a current slot, the estimated channel correlation indicating a rate of change of a wireless channel over time; and a Doppler spread estimator configured to estimate a Doppler spread of the wireless channel by supplying the features to one or more Doppler shift predictors trained on training data across a training signal-to-noise ratio (SNR) range and across a training Doppler shift range, each Doppler shift predictor being trained on a portion of the training data corresponding to a different portion of the training data.