Parallel-Input FIR Filter for Flexible Decimation at High Sample Rates
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current finite impulse response (FIR) filters face challenges in accommodating high sample rates and flexible decimation factors, particularly in hardware implementations, where multipliers consume significant resources and polyphase filters are inefficient with high decimation factors and long filter lengths.
Innovation Solution
A decimating transpose form FIR filter design that receives parallel signal samples, multiplies them by filter coefficients, and uses a summing tree to differentially delay and sum partial sums, allowing for flexible decimation factors and reduced clock rates for multipliers, along with a coefficient memory for selecting filter coefficients based on the decimation factor.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If polyphase filter is used to achieve flexible decimation factors, then decimation flexibility is improved, but the number of multipliers increases significantly
Solution Approach 1:
The filter is divided into multiple parallel sub-filters, each processing a different phase of the input signal. This segmentation allows the system to achieve flexible decimation factors while reducing the computational burden on each individual multiplier, as each multiplier only needs to process one coefficient per input sample rather than multiple coefficients.
Solution Approach 2:
The patent transitions from a time-sequential processing approach to a parallel spatial processing approach by implementing multiple sub-filters that operate simultaneously on different phases of the input signal. This dimensional change from sequential to parallel processing enables flexible decimation without proportionally increasing the number of multipliers.
2Speed
If high sample rates are accommodated by using traditional FIR filter structures, then sample rate capability is improved, but resource consumption increases
Solution Approach 1:
The high sample rate processing is divided into multiple parallel sub-filters that each operate at a lower effective rate. By segmenting the processing workload across multiple sub-filters working in parallel, the system can accommodate high input sample rates without requiring each individual multiplier to operate at the full sample rate, thus reducing resource consumption.
Solution Approach 2:
Multiple sub-filters are merged into a single parallel processing structure where their outputs are combined. This merging allows the system to achieve high sample rate capability through parallel processing while each individual multiplier unit consumes fewer resources, as the total computational workload is distributed across the merged sub-filters.
3Productivity
If decimation is performed by keeping only some output samples, then output data rate is reduced, but the number of computations that can be discarded increases
Solution Approach 1:
The filter structure is pre-configured with multiple sub-filters that compute only the specific output samples that will be retained after decimation. By performing preliminary organization of the filter structure to match the decimation pattern, computations for samples that will be discarded are avoided entirely, reducing wasted energy while maintaining the desired output data rate.
Data Source
AI summary
A filter and method for finite impulse response filtering an input signal is described. In one embodiment, the filter includes an input circuit configured to receive input samples in parallel, where the parallel input samples correspond to sequential samples of the input signal. In another embodiment, the filter includes a coefficient memory configured to store filter response coefficients and output a subset of those coefficients corresponding to a selected decimation factor.