QC-LDPC Decoder Layer Scheduling Using Orthogonal H Matrix Rows
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current quasi-cyclic low density parity check (QC-LDPC) decoders experience stall periods due to inherent delays, limiting their throughput as they wait for soft information elements to be updated before processing subsequent layers.
Innovation Solution
The method involves generating a parity-check matrix (H matrix) with constrained rows, where each matrix value is associated with initial soft information elements that exclude a specific number of preceding elements, reducing the stall period by allowing immediate processing of layers after a shorter delay (D-E) cycles, rather than the full inherent delay (D) cycles.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the decoder waits for the full inherent delay (D) cycles to update soft information elements before processing subsequent layers, then the accuracy of decoded information is improved, but the throughput is reduced due to stall periods
Solution Approach 1:
The patent applies partial action by updating only the specific soft information elements that are ready (those not within the excluded range of E*P preceding elements) rather than waiting for all elements to be fully updated. This allows the decoder to process subsequent layers with partially updated information, reducing stall periods while maintaining acceptable decoding accuracy.
Solution Approach 2:
The patent implements preliminary action by pre-calculating and storing the H matrix with constrained rows that identify which soft information elements are ready for processing. This allows the decoder to immediately proceed with layer processing without waiting for the full inherent delay, as the readiness of soft information elements is determined in advance through the constrained H matrix structure.
2Productivity
If the decoder processes subsequent layers immediately after a shorter delay (D-E) cycles, then the throughput is improved, but the stall period reduction may affect decoding accuracy
Solution Approach 1:
The patent applies local quality by differentiating the processing timing for different soft information elements based on their readiness status. Specifically, elements within the excluded range of E*P preceding elements are handled differently (wait for full update) versus other elements (can be processed immediately). This localized differentiation allows throughput improvement while preserving accuracy for critical elements.
Solution Approach 2:
The patent changes the timing parameter from a fixed inherent delay (D) to a variable delay (D-E) based on the readiness of soft information elements. The parameter E represents the number of cycles by which the stall period is reduced, and this parameter is optimized to balance throughput improvement with maintaining sufficient update time for accurate decoding.
3Productivity
If the H matrix is designed with constrained rows to reduce stall periods, then the throughput is enhanced, but the matrix design complexity increases
Solution Approach 1:
The patent segments the H matrix into constrained rows that specifically identify which soft information elements are ready for processing. This segmentation allows the decoder to systematically determine which elements can be processed immediately versus which require waiting, providing a structured approach to reducing stall periods while maintaining manageable design complexity through modular matrix construction.
Data Source
AI summary
A method and apparatus for a quasi-cyclic low density parity check (QC-LDPC) decoder utilizes a parity check matrix (H matrix) having a matrix value for each row and column position in the matrix. Each matrix value is associated with an initial soft information element where, for each one of the matrix values associated with a constrained row, the one of the matrix values is constrained to a set of constraint values associated with a set of initial soft information elements. The set of initial soft information elements excludes a number of soft information elements that immediately precede a first initial soft information element. The first initial soft information element is associated with a selected first matrix value associated with a first row that immediately precedes the constrained row, and with the same column as the one of the matrix values in the constrained row.