Kernel Adaptive Filter for Non-Linear Self-Interference Cancellation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Adaptive filters used for digital self-interference cancellation in full duplex communication systems are ineffective when non-linear components like power amplifiers are present, as they rely on linear channel models, leading to decreased cancellation performance.
Innovation Solution
The implementation of a kernel adaptive filter with a new kernel design, specifically using a kernel function κ(x,y) = real(xy*)exp(β|xy*|, that maps input signals to a potentially infinite dimension, allowing for better modeling of non-linear distortions and improved interference cancellation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If adaptive filters with linear channel models are used for digital self-interference cancellation, then the system structure remains simple, but cancellation performance severely decreases when non-linear components such as power amplifiers are present
Solution Approach 1:
The patent transforms the linear filter parameters into kernel parameters that can capture non-linear relationships. By using a kernel function κ(x,y) = real(xy*)exp(β|xy*|), the filter adapts to non-linear channel characteristics while maintaining the adaptive filtering framework, thus improving cancellation performance in the presence of power amplifiers without completely redesigning the system architecture
Solution Approach 2:
The kernel adaptive filter maps the input signals to a potentially infinite dimension space, allowing the filter to model non-linear distortions by operating in a higher-dimensional feature space. This dimensional transformation enables the filter to capture complex non-linear relationships that cannot be represented in the original signal space
2Reliability
If kernel adaptive filter with complex kernel function is used to model non-linear distortions, then cancellation performance improves, but computational complexity increases
Solution Approach 1:
The patent introduces a single real parameter β in the kernel function that controls the non-linear modeling capability. By adjusting this parameter, the system can adapt to different non-linear channel conditions while maintaining computational efficiency. The kernel function κ(x,y) = real(xy*)exp(β|xy*|) provides a flexible yet computationally manageable way to model non-linear distortions
Data Source
AI summary
The disclosure relates to a cancellation device for cancelling interference caused by a non-linear device, the cancellation device comprising: an adaptive filter coupled in parallel to the non-linear device, wherein the adaptive filter is configured to filter an input signal of the non-linear device to generate an approximation signal approximating an output signal of the non-linear device; an error signal generator configured to generate an error signal based on a function of the output signal and the approximation signal; and a controller configured to adjust the adaptive filter based on the error signal and a phase and magnitude relation of the output signal and the input signal.