LDPC Encoder Matrix Reconstruction for Low-Storage Encoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Storing large parity check matrices for LDPC codes is costly and inefficient, requiring significant hardware resources without effectively minimizing storage requirements.
Innovation Solution
Indirect storage of non-regular LDPC matrix portions using a look-up table to store indices and distance/location parameters, allowing for efficient retrieval of matrix elements through address generation and circular shift operations, thereby reducing storage needs without compromising encoding performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If large parity check matrices are stored directly, then encoding performance is maintained, but storage requirements and hardware complexity increase significantly
Solution Approach 1:
The patent uses a look-up table to store only the indices and distance parameters that define the parity check matrix structure, rather than storing the complete matrix. The actual matrix elements are generated on-demand by copying and shifting base patterns according to the stored parameters, dramatically reducing storage requirements while maintaining encoding performance.
Solution Approach 2:
The patent pre-calculates and stores only the essential parameters (indices and distance values) that define the matrix structure in advance. These preliminary parameters are sufficient to reconstruct any element of the parity check matrix when needed, avoiding the need to store all matrix elements directly.
2Measurement precision
If complete parity check matrices are stored, then encoding accuracy is preserved, but hardware resources and cost increase
Solution Approach 1:
Instead of storing complete matrix data that would require large hardware memory resources, the patent stores compact index and distance parameters in a look-up table. The actual matrix elements are generated by copying and transforming base patterns using these parameters, preserving encoding accuracy while minimizing hardware resources.
Solution Approach 2:
The patent extracts only the essential defining parameters (indices and distances) from the complete parity check matrix and stores them in the look-up table. The redundant information is eliminated, and only the minimal necessary data is retained to reconstruct matrix elements when needed for encoding operations.
3Ease of operation
If traditional storage methods are used, then matrix elements are readily accessible, but storage efficiency decreases
Solution Approach 1:
The patent stores compact index and distance parameters in the look-up table that can be quickly accessed. When matrix elements are needed, they are generated by copying and shifting predefined base patterns according to these parameters, achieving both fast accessibility and high storage efficiency.
Solution Approach 2:
The look-up table acts as an intermediary between the stored parameters and the required matrix elements. It provides the index and distance values that mediate the transformation from compact parameter storage to full matrix element generation, enabling efficient access without storing complete matrices.
Data Source
AI summary
A method for encoding digital data with a low-density parity check (LDPC) matrix includes: indirectly storing a non-regular portion of the LDPC matrix by storing a plurality of indices corresponding to a plurality of non-zero sub-matrices of the non-regular portion, and by storing a plurality of distance/location parameters respectively corresponding to numbers of zero sub-matrices between adjacent non-zero sub-matrices of the non-regular portion or respectively corresponding to distances between adjacent non-zero sub-matrices of the non-regular portion; generating at least one address according to at least one distance/location parameter; accessing information bits corresponding to the address; and recovering at least one element of the LDPC matrix according to at least one index and the information bits to encode the digital data according to the LDPC matrix.