Arithmetic-Only Adaptive Filter for Frequency-Domain Signal Processing

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Wireline communication devices require complex circuitry for frequency-domain filters, leading to large size and high power consumption, especially when filtering entire channels.

Innovation Solution

Implementing adaptive filters using arithmetic-only operations to approximate windowing functions, reducing the need for complex matrix operations like Fast Fourier Transform and Inverse Fast Fourier Transform, and using a sum of sinusoidal functions to approximate windowing functions.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If complex circuitry for Fast Fourier Transform and Inverse Fast Fourier Transform is used to implement frequency-domain filters, then filtering effectiveness is improved, but device area and power consumption increase significantly

Engineering Contradiction:
Improvefiltering effectivenessVSAvoiddevice area
Core Design Contradiction:
ReliabilityVSArea of stationary object

Solution Approach 1:

The patent replaces complex FFT/IFFT circuitry with arithmetic operations in the frequency domain. Specifically, it substitutes the mechanical/computational complexity of transform operations with direct arithmetic computations (multiplications and additions) operating on frequency-domain samples, thereby reducing device area while maintaining filtering effectiveness

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The patent changes the operational parameters by working directly in the frequency domain without performing FFT/IFFT transformations. By modifying the approach from time-domain transformation to direct frequency-domain arithmetic operations, it achieves the same filtering effect with reduced computational complexity and smaller device area

Inventive Principle:
Principle #35Parameter changes

2Reliability

If complex circuitry for Fast Fourier Transform and Inverse Fast Fourier Transform is used to implement frequency-domain filters, then filtering effectiveness is improved, but power consumption increases significantly

Engineering Contradiction:
Improvefiltering effectivenessVSAvoidpower consumption
Core Design Contradiction:
ReliabilityVSUse of energy by stationary object

Solution Approach 1:

The patent replaces power-intensive FFT/IFFT circuitry with simpler arithmetic operations in the frequency domain. By substituting complex transform computations with direct arithmetic operations (multiplications and additions), it significantly reduces power consumption while preserving filtering effectiveness

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The patent extracts and eliminates the unnecessary FFT/IFFT transformation steps from the filtering process. By removing these computationally expensive operations and working directly in the frequency domain with arithmetic operations, it reduces power consumption while maintaining the essential filtering functionality

Inventive Principle:
Principle #2Taking out (Extraction)

3Measurement precision

If windowing functions are implemented using complex matrix operations, then filtering accuracy is improved, but device complexity increases

Engineering Contradiction:
Improvefiltering accuracyVSAvoidcircuit complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent replaces complex matrix operations for windowing with simple arithmetic operations in the frequency domain. By substituting matrix multiplications and transformations with direct arithmetic computations on frequency-domain samples, it reduces device complexity while maintaining filtering accuracy

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The patent inverts the conventional approach by performing windowing operations directly in the frequency domain rather than transforming to time domain, applying the window, and transforming back. This reverse approach eliminates complex matrix operations while preserving the accuracy benefits of windowing

Inventive Principle:
Principle #13The other way round (Inversion)

Data Source

PatentUS11632147B2Simplified frequency-domain filter adaptation window
Publication Date: 2023.04.18 MARVELL ASIA PTE LTD
  • US11632147B2 patent drawing
  • US11632147B2 patent drawing
  • US11632147B2 patent drawing

AI summary

A physical layer transceiver for a data channel includes receiver circuitry configured to receive signals on the data channel, transmit circuitry configured to transmit signals onto the data channel, and adaptive filter circuitry coupled to the receiver circuitry and the transmit circuitry and configured to filter the data channel by operating on input frequency-domain data samples to output filtered data samples. The adaptive filter circuitry includes error sample generation circuitry configured to generate error samples representing a difference between a target response and the filtered data samples, arithmetic-only circuitry configured to approximate a windowing function to operate on the error samples, and output sample generation circuitry configured to operate on windowed error samples to provide the output filtered data samples. The comparison circuitry may be configured for time-domain operation and may further be configured to transform the error signals into frequency-domain error signals.