Non-Binary Check Node Processing with Pre-Sorted Message Truncation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

The high computational complexity and memory requirements of check node processing units in non-binary linear block error correcting codes make hardware implementations expensive and inefficient, particularly for long codes or large Galois fields, limiting their practical application in digital communication systems.

Innovation Solution

The implementation of a check node processing unit that reduces the number of input messages required to compute output messages, using methods such as sorting and truncation of messages, and employing architectures like syndrome decoding or forward-backward processing to minimize computational complexity and silicon area, thereby improving throughput and reducing costs.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If check node processing units use traditional algorithms for non-binary linear block error correcting codes, then error correction capability is maintained, but computational complexity and memory requirements become excessively high

Engineering Contradiction:
Improveerror correction capabilityVSAvoidcomputational complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The check node processing is divided into multiple stages: input message reception, sorting of input messages according to reliability metrics, selective processing of only certain input messages based on sorting results, and output message generation. This segmentation allows the system to process only the most relevant messages, reducing computational complexity while maintaining error correction capability.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

Instead of processing all input messages equally, the algorithm performs partial processing by selectively handling only the most reliable input messages (those with highest reliability metrics). This partial action approach reduces the number of computations required while still achieving sufficient error correction performance.

Inventive Principle:
Principle #16Partial or excessive action

2Measurement precision

If check node processing units process all input messages to ensure accurate output messages, then decoding accuracy is maintained, but silicon area and processing time increase significantly

Engineering Contradiction:
Improvedecoding accuracyVSAvoidthroughput
Core Design Contradiction:
Measurement precisionVSProductivity

Solution Approach 1:

Input messages are sorted according to their reliability metrics before processing. This preliminary sorting action identifies which messages are most likely to contribute to accurate output messages, allowing the system to skip processing of less reliable messages and thereby increase throughput while maintaining decoding accuracy.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The processing intensity is varied locally based on the quality (reliability) of input messages. High-reliability messages receive full processing attention, while lower-reliability messages are either processed with reduced complexity or skipped entirely. This local quality approach optimizes the balance between accuracy and throughput.

Inventive Principle:
Principle #3Local quality

3Reliability

If check node processing units implement full computational algorithms for non-binary codes, then error correction performance is optimized, but hardware cost and implementation complexity increase

Engineering Contradiction:
Improveerror correction performanceVSAvoidhardware implementation cost
Core Design Contradiction:
ReliabilityVSEase of manufacture

Solution Approach 1:

The algorithm extracts and processes only the essential components of input messages (those with highest reliability metrics) while discarding or skipping less critical processing steps. This extraction approach reduces hardware requirements and implementation cost while maintaining sufficient error correction performance.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The processing approach changes based on parameters such as message reliability metrics, code length, and field size. By adapting the level of processing to these parameters, the system achieves good error correction performance without always requiring the full computational overhead, thereby reducing hardware implementation cost.

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP3242405B1Non-binary check node processing with pre-sorted input
Publication Date: 2024.02.21 UNIVERSITY OF SOUTHERN BRITTANY
  • EP3242405B1 patent drawingFigure 1~2
  • EP3242405B1 patent drawingFigure 3~5
  • EP3242405B1 patent drawingFigure 6

AI summary

Embodiments of the invention provide a check node processing unit implemented in a decoder, said decoder being configured to decode a signal encoded using an error correcting code, said signal comprising symbols, the check node processing unit being configured to receive at least two input messages and to generate at least one output message, each message comprising a plurality of components, each component comprising a value of a symbol and a reliability metric associated with said symbol, wherein the check node processing unit comprises: - a data structure (31) configured to store said input messages, the components of the input messages being associated with an integer index in the data structure; - a data processing unit (33) configured to apply one or more iterations of a transformation operation to at least a part of the data structure, each iteration of the transformation operation being performed to arrange the components of said input messages in said data structure (31) depending on at least some of the components of the messages associated with a given value of the integer index, which provides a transformed data structure; - a calculation unit (35) configured to determine said at least one output message from the components comprised in said transformed data structure.