Volterra-Feature DNN Equalizer for Nonlinear PA Distortion

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current power amplifier (PA) technologies face challenges in achieving high power efficiency due to high peak-to-average power ratio (PAPR) in modern wireless communication systems, particularly in 5G systems, leading to severe reduction in efficiency and increased costs for devices like those in massive machine-type communications and IoT, where existing nonlinear PA mitigation strategies are insufficient.

Innovation Solution

A deep neural network (DNN)-based equalizer is implemented to mitigate PA distortions by exploiting Volterra series nonlinearity modeling, constructing input features for rapid convergence and efficient nonlinear response, integrating with neural networks for low-cost and high-performance PA equalization, surpassing conventional digital predistorters.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If digital pre-distorter (DPD) is used to linearize PAs, then nonlinear distortion compensation capability is improved, but device complexity and implementation cost increase

Engineering Contradiction:

Improvenonlinear distortion compensation capabilityVSAvoidimplementation complexity

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The patent combines Volterra series modeling with deep neural networks to create a unified equalization framework. The Volterra series provides the nonlinear modeling foundation while the DNN performs the equalization function, merging two separate techniques into a single integrated system that achieves better distortion compensation than conventional DPD with reduced complexity

Inventive Principle:

Principle #5Merging (Combining)

Solution Approach 2:

The patent replaces the traditional DPD mechanical/architectural approach with a data-driven deep neural network system. Instead of using complex DPD structures with multiple feedback paths and adaptive algorithms, the invention uses a DNN that learns the inverse nonlinear characteristics directly from training data, substituting a simpler computational approach for the complex mechanical-like DPD structure

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

2Reliability

If high output backoff (OBO) is used to suppress nonlinear distortions, then distortion is reduced, but power efficiency deteriorates

Engineering Contradiction:

Improvedistortion suppressionVSAvoidpower efficiency

Core Design Contradiction:

ReliabilityVSUse of energy by moving object

Solution Approach 1:

The patent applies preliminary equalization at the receiver side before signal processing. By pre-compensating for the nonlinear distortions using the DNN equalizer, the system can operate the PA at high efficiency points without requiring high OBO, as the distortion compensation is handled by the learned equalization model rather than by reducing PA backoff

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent changes the operational parameters of the system by using a data-driven equalization approach that allows the PA to operate in its high-efficiency nonlinear region. The DNN is trained to compensate for various PA operating conditions, enabling the system to maintain distortion suppression while operating at parameters that maximize power efficiency rather than requiring conservative backoff settings

Inventive Principle:

Principle #35Parameter changes

3Measurement precision

If Volterra series model is used for nonlinear PA modeling, then modeling accuracy is improved, but computational complexity increases

Engineering Contradiction:

Improvenonlinear modeling accuracyVSAvoidcomputational complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent extracts the essential nonlinear characteristics from the Volterra series model and uses them as input features for the DNN. Instead of implementing the full Volterra series computation which is computationally intensive, the invention extracts the key nonlinear modeling insights and feeds them into a more efficient neural network structure, separating the modeling function from the computation function

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

The patent introduces the DNN as an intermediary between the Volterra series model and the equalization output. The Volterra series provides the theoretical foundation and input feature construction, while the DNN acts as an intermediary computational engine that performs the actual equalization more efficiently than direct Volterra series inversion, bridging the gap between accurate modeling and practical implementation

Inventive Principle:

Principle #24Intermediary (Mediator)

Data Source

PatentUS10985951B2Integrating Volterra series model and deep neural networks to equalize nonlinear power amplifiers

Publication Date: 2021.04.20 THE RES FOUNDATION FOR THE STATE UNIV OF NEW YORK

AI summary

The nonlinearity of power amplifiers (PAs) has been a severe constraint in performance of modern wireless transceivers. This problem is even more challenging for the fifth generation (5G) cellular system since 5G signals have extremely high peak to average power ratio. Nonlinear equalizers that exploit both deep neural networks (DNNs) and Volterra series models are provided to mitigate PA nonlinear distortions. The DNN equalizer architecture consists of multiple convolutional layers. The input features are designed according to the Volterra series model of nonlinear PAs. This enables the DNN equalizer to effectively mitigate nonlinear PA distortions while avoiding over-fitting under limited training data. The non-linear equalizers demonstrate superior performance over conventional nonlinear equalization approaches.