Deep Learning OFDM Receiver for One-Bit Quantization Distortion

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Low resolution quantization in wireless receivers, particularly in OFDM systems, leads to severe inter-carrier interference and poor performance at medium to high signal-to-noise ratios, limiting their effectiveness in high-data-rate wireless systems.

Innovation Solution

The implementation of deep learning-based architectures, including generative supervised deep neural networks for channel estimation and autoencoders for data detection, which utilize low resolution analog-to-digital converters to manage distortion and improve performance under low resolution quantization conditions.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Use of energy by stationary object

If low resolution quantization is used in ADCs, then power consumption and hardware complexity are reduced, but measurement precision and performance at medium to high SNRs deteriorate

Engineering Contradiction:

Improvepower consumptionVSAvoidquantization precision

Core Design Contradiction:

Use of energy by stationary objectVSMeasurement precision

Solution Approach 1:

A neural network is introduced as an intermediary component between the low-resolution ADC and the signal processing chain. The neural network learns to compensate for the quantization distortion introduced by the low-resolution ADC, effectively acting as a mediator that restores signal quality without requiring high-resolution hardware

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The system changes the resolution parameter of the ADC from high (e.g., 12-bit) to low (e.g., 1-6 bit) to reduce power consumption, while simultaneously using a neural network to adaptively adjust and compensate for the resulting quantization effects, thereby maintaining acceptable performance

Inventive Principle:

Principle #35Parameter changes

2Device complexity

If low resolution quantization is used, then device complexity is reduced, but reliability and performance in OFDM systems deteriorate due to severe inter-carrier interference

Engineering Contradiction:

ImproveADC complexityVSAvoidOFDM system performance

Core Design Contradiction:

Device complexityVSReliability

Solution Approach 1:

The neural network serves as an intermediary that compensates for the inter-carrier interference introduced by low-resolution quantization in OFDM systems, restoring the orthogonality between subcarriers that is essential for reliable OFDM operation

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent replaces the traditional mechanical/hardware approach of using high-resolution ADCs with a software-based neural network approach that learns to compensate for quantization effects, thereby maintaining OFDM performance with simpler hardware

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

3Measurement precision

If high resolution ADCs are used, then measurement precision is improved, but power consumption and hardware complexity increase

Engineering Contradiction:

Improvequantization precisionVSAvoidpower consumption

Core Design Contradiction:

Measurement precisionVSUse of energy by stationary object

Solution Approach 1:

Instead of using high-resolution ADCs to directly achieve precise measurements, the patent inverts the approach by using low-resolution ADCs combined with a neural network that learns to reconstruct the high-resolution signal, thereby achieving precision through software rather than hardware

Inventive Principle:

Principle #13The other way round (Inversion)

4Device complexity

If one-bit quantization is used, then device complexity is minimized, but measurement precision and reliability deteriorate severely

Engineering Contradiction:

ImproveADC complexityVSAvoidsignal reconstruction accuracy

Core Design Contradiction:

Device complexityVSMeasurement precision

Solution Approach 1:

The neural network acts as a powerful intermediary that compensates for the extreme quantization distortion introduced by one-bit ADCs, learning to reconstruct the original high-resolution signal from the severely quantized one-bit samples

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The system creates a composite architecture combining one-bit ADC hardware with a neural network software component, where the combination achieves performance that neither component could achieve alone, analogous to composite materials having properties superior to their individual constituents

Inventive Principle:

Principle #40Composite materials

Data Source

PatentUS12126467B2Low resolution OFDM receivers via deep learning

Publication Date: 2024.10.22 BOARD OF RGT THE UNIV OF TEXAS SYST

AI summary

Various embodiments provide for deep learning-based architectures and design methodologies for an orthogonal frequency division multiplexing (OFDM) receiver under the constraint of one-bit complex quantization. Single bit quantization greatly reduces complexity and power consumption in the receivers, but makes accurate channel estimation and data detection difficult. This is particularly true for OFDM waveforms, which have high peak-to average (signal power) ratio in the time domain and fragile subcarrier orthogonality in the frequency domain. The severe distortion for one-bit quantization typically results in an error floor even at moderately low signal-to-noise-ratio (SNR) such as 5 dB. For channel estimation (using pilots), various embodiments use novel generative supervised deep neural networks (DNNs) that can be trained with a reasonable number of pilots. After channel estimation, a neural network-based receiver specifically, an autoencoder jointly learns a precoder and decoder for data symbol detection.