Parallel IIR Filter Segmentation for High-Speed Serial Data Throughput
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for high-bitrate serial data transfer, such as those using Tomlinson-Harashima precoding filters, face limitations in clock rates due to CMOS process constraints and inefficiencies in parallelization, leading to suboptimal bit error rates and throughput.
Innovation Solution
The solution involves combining data signals with reference signals before filtering, allowing for equal or comparable system responses from IIR filters, which enables the multiplexing of filtered data streams as if they were processed by a single filter, reducing settling time and increasing throughput by parallelizing channel coding and reducing filter power dissipation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If parallelization of THP filters is increased to increase throughput, then clock rate is improved, but efficiency decreases due to longer feedback paths
Solution Approach 1:
The patent divides the single long feedback path into multiple shorter feedback paths by segmenting the THP filter into parallel filter structures. Each filter processes a portion of the data stream independently with shorter feedback delays, enabling higher clock rates while maintaining overall throughput efficiency through parallel processing.
Solution Approach 2:
The patent transitions from sequential single-filter processing to parallel multi-filter processing, adding a temporal dimension to the signal processing architecture. Multiple filters operate simultaneously on different portions of the data stream, converting a single long feedback path into multiple shorter paths that can be processed in parallel.
2Productivity
If continuous parallel processing is implemented to increase throughput, then data transfer rate is improved, but filter power dissipation increases
Solution Approach 1:
The patent segments the data stream into multiple portions that can be processed in parallel by multiple filters. Each filter processes a subset of the data with lower individual complexity, reducing the power dissipation per filter while maintaining high overall throughput through parallel processing of multiple data portions simultaneously.
3Device complexity
If single filter processes all data sequentially, then filter complexity is reduced, but settling time increases and throughput is limited
Solution Approach 1:
The patent divides the data stream into multiple portions that can be processed simultaneously by parallel filters. This segmentation allows the system to achieve high throughput by processing multiple data portions in parallel while keeping individual filter complexity manageable through the distributed processing architecture.
Data Source
AI summary
A transmitter for transmitting an output signal includes first and second filter structures. The first filter structure includes a first combiner to extend a first data signal by a first reference signal to obtain a first extended data signal, and a first IIR filter for filtering the first extended data signal to obtain a first filtered data signal. The second filter structure includes a second combiner to extend a second data signal by a second reference signal, and a second IIR filter for filtering the second extended data signal. The transmitter includes a multiplexer for combining the first and second filtered data signals to obtain the output signal. A system response of the first IIR filter based on the first reference signal corresponds to a system response of the second IIR filter based on the second reference signal.