Primal-Dual Tanner Graph Decoding for Short Error-Correcting Codes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current error-correcting codes, such as turbo-codes and LDPC codes, are inefficient for short lengths due to sub-optimal decoding algorithms and high hardware costs, particularly for codes with small to medium lengths (n < 1000), as they struggle with dense generating matrices and complex decoding processes.
Innovation Solution
A method that simultaneously uses a primal error-correcting code and its dual form to reduce decoding iterations and hardware complexity, employing a primal-dual graph structure that eliminates cycles and optimizes information propagation, allowing for real-time decoding of short length codes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If turbo-codes or LDPC codes are used for error correction, then error correction performance is improved for large code lengths (n > 1000), but decoding complexity and hardware cost increase significantly
Solution Approach 1:
The patent segments the decoding process into two independent parts: a first decoder handling the primal code and a second decoder handling the dual code. Each decoder operates separately on its own code structure, allowing independent optimization and reducing the overall complexity compared to a single unified decoder for the same code length.
Solution Approach 2:
The patent merges two different coding approaches (primal code and dual code) into a single error correction system. By combining the strengths of both codes and using parallel decoding structures, the system achieves better error correction performance than either code alone while maintaining manageable complexity through modular architecture.
2Productivity
If probability propagation algorithm is used for decoding, then decoding speed is improved, but performance deteriorates for codes with small minimum length cycles
Solution Approach 1:
The patent segments the error correction functionality into two separate codes with different cycle structures. The primal code and dual code each have complementary properties, where the weaknesses of one (short cycles) are compensated by the strengths of the other, allowing probability propagation to work effectively on both structures simultaneously.
Solution Approach 2:
The patent creates a composite error correction system by combining two different codes (primal and dual) with complementary characteristics. This composite approach leverages the advantageous properties of both codes to achieve robust error correction across various code lengths and cycle structures.
3Measurement precision
If exhaustive decoding algorithm is used, then decoding accuracy is improved, but computational complexity becomes exponential O(2^k)
Solution Approach 1:
The patent divides the exponentially complex exhaustive decoding problem into two separate, more manageable decoding tasks. Each decoder handles one code (primal or dual) independently, reducing the computational burden from exponential to polynomial complexity for each segment, while maintaining high accuracy through the complementary nature of the two codes.
Data Source
Figure 1~2C
Figure 3A~3C
Figure 4~6B
AI summary
The invention relates to methods and devices for coding and decoding coded data comprising source data (S) and redundancy data (R), said redundancy data (R) being obtained by applying, upon coding, an error corrector code to said source data, implementing a generating matrix comprising, in its systematic form, an identity matrix and an invertible matrix for switching from the source data to the redundancy data (P), the coding or decoding (12) being based on a Tanner graph which combines the graph of said error corrector code and the graph of the dual code of said error corrector code by superimposing these graphs.