Partitioned FIR Filtering for Low-Latency Frequency-Domain Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing signal processing systems face challenges with high memory requirements and processor load due to the use of FIR filters, stability issues with IIR filters, and high latency in multirate digital systems, particularly in applications like hands-free voice communication where echo cancellation and signal enhancement are critical.
Innovation Solution
A signal processing system employing time and frequency domain partitioning, using a converter to generate Fourier transformed signal partitions, which are then convoluted with Fourier transformed filter partitions to produce a processed digital output signal, thereby reducing computational complexity and memory requirements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If FIR filters are used for signal processing, then stability is improved, but memory requirements and processor load increase significantly
Solution Approach 1:
The patent divides the filter processing into multiple partitions in both time and frequency domains. The signal is segmented into frequency sub-bands using FFT, and each sub-band is processed independently with smaller filter coefficients, reducing overall memory requirements while maintaining stability.
Solution Approach 2:
The patent transforms the filter operation from purely time-domain to frequency-domain using Fast Fourier Transform (FFT). By converting signals to frequency domain, the convolution operation becomes simple multiplication, dramatically reducing computational complexity and memory access requirements.
2Reliability
If FIR filters are used for signal processing, then stability is improved, but processor load increases significantly
Solution Approach 1:
The patent transforms the filter operation from purely time-domain to frequency-domain using Fast Fourier Transform (FFT). By converting signals to frequency domain, the convolution operation becomes simple multiplication, dramatically reducing computational complexity and memory access requirements.
Solution Approach 2:
The patent pre-computes and stores the FFT of filter coefficients in frequency domain. This preliminary action allows the actual filtering operation to proceed with simple complex multiplications rather than lengthy convolution sums, reducing real-time processor load.
3Device complexity
If multirate digital systems are used for signal processing, then computational complexity is reduced, but latency increases
Solution Approach 1:
The patent uses overlapping windows with dynamic weighting functions (e.g., Hann, Hamming) to smoothly transition between adjacent signal blocks. This dynamic approach eliminates discontinuities at block boundaries without requiring long processing windows, thus reducing latency while maintaining low computational complexity.
Solution Approach 2:
The patent employs periodic application of window functions with carefully chosen overlap lengths (typically 50% overlap). This periodic structure allows continuous processing with reduced block sizes, lowering latency while maintaining computational efficiency through reusable window functions.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach enables efficient signal processing with reduced latency and memory usage, effectively enhancing speech signal quality and reducing echo/feedback issues in communication systems.
Implementation Method 1
A converter is used to convert the digital input signal to provide a first plurality of Fourier transformed signal partitions
Implementation Method 2
The partitions of the converter and the filter signal source are provided to a convolution processor that uses the partitions to generate a plurality of convoluted partitioned output signals
Data Source
AI summary
The present invention relates to a method for processing a digital input signal by a Finite Impulse Response, FIR, filtering means, comprising partitioning the digital input signal at least partly in the time domain to obtain at least two partitions of the digital input signal; partitioning the FIR filtering means in the time domain to obtain at least two partitions of the FIR filtering means; Fourier transforming each of the at least two partitions of the digital input signal to obtain Fourier transformed signal partitions; Fourier transforming each of the at least two partitions of the FIR filtering means to obtain Fourier transformed filter partitions; performing a convolution of the Fourier transformed signal partitions and the corresponding Fourier transformed filter partitions to obtain spectral partitions; combining the spectral partitions to obtain a total spectrum; and inverse Fourier transforming the total spectrum to obtain a digital output signal.