LDPC Puncturing Using EMD to Preserve Decoding Performance
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for puncturing Low Density Parity Check (LDPC) codes face limitations in supporting various code rates and minimizing performance degradation, particularly for regular and irregular codes, and are not applicable to all types of LDPC codes due to complexities in parity check matrix design and puncturing position determination.
Innovation Solution
A method for puncturing LDPC codes that involves setting a codeword length and selecting check nodes and bit nodes based on priority, using Extrinsic Message Degree (EMD) to minimize the number of punctured bits connected to each check node, and ensuring uniform puncturing counter values to prevent performance degradation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If puncturing is performed to vary code rate, then code rate flexibility is improved, but decoding performance deteriorates
Solution Approach 1:
The patent applies local quality by differentiating puncturing strategies for systematic bits versus parity bits. Systematic bits are punctured based on their connection to check nodes with lowest EMD values, while parity bits are punctured separately. This localized differentiation ensures that puncturing occurs in regions least critical for decoding performance, thus maintaining reliability while achieving code rate flexibility.
Solution Approach 2:
The patent changes the parameter of puncturing position selection by using EMD (Extrinsic Message Degree) as a criterion. Instead of random or uniform puncturing, bits connected to check nodes with lowest EMD values are selected for puncturing. This parameter-based selection optimizes which bits are punctured, minimizing performance degradation while enabling variable code rates.
2Manufacturing precision
If puncturing position is designated by degree distribution, then puncturing optimization is improved, but applicability to regular codes deteriorates
Solution Approach 1:
The patent achieves universality by designing a puncturing method that works for both regular and irregular LDPC codes. Instead of relying on degree distribution specific to irregular codes, the patent uses EMD values which can be calculated for any LDPC code structure. This universal approach allows the same puncturing strategy to be applied across different code types, maintaining optimization while expanding applicability.
Solution Approach 2:
The patent inverts the conventional approach by not selecting bits based on highest degree distribution, but rather by identifying check nodes with lowest EMD values and selecting bits connected to them. This inversion of the selection criterion enables the method to work effectively for regular codes where degree distribution is uniform, thus expanding applicability while maintaining puncturing optimization.
3Reliability
If maximum number of punctured bits per check node is minimized, then decoding convergence is improved, but puncturing flexibility deteriorates
Solution Approach 1:
The patent applies dynamics by making the puncturing pattern adaptive rather than fixed. The EMD values are calculated based on the specific LDPC code structure and desired code rate, allowing the puncturing pattern to dynamically adjust to different scenarios. This dynamic approach ensures that the maximum number of punctured bits per check node is minimized for decoding convergence, while simultaneously providing flexibility to achieve various code rates.
Solution Approach 2:
The patent uses feedback by calculating EMD values that reflect the impact of puncturing on decoding performance. These EMD values provide feedback information about which check nodes are most affected by puncturing, allowing the algorithm to adjust puncturing positions to minimize the maximum number of punctured bits per check node. This feedback mechanism maintains decoding convergence while preserving puncturing flexibility for different code rates.
Data Source
AI summary
A method for puncturing a Low Density Parity Check (LDPC). The method includes a) setting a codeword length and the total number of bit nodes to be punctured; b) selecting a check node (or check nodes) with highest priority excluding check nodes completely checked in a current round; c) selecting a bit node (or bit nodes) with a highest priority excluding bit nodes completely checked among bit nodes connected to the selected check node (or check nodes); d) determining whether the selected bit node is a bit node to be punctured, that is, it is not systematic, not set by a puncturing prohibition flag; e) puncturing an associated bit node if the selected bit node is the bit node to be punctured, setting unpunctured bit nodes connected to the selected check node by a puncturing prohibition flag, decreasing the number of remained bit nodes to be punctured by 1 and increasing the number of connected punctured node of associated check node by 1; f) determining whether the number of remaining bits to be punctured is greater than 0; and g) returning to step b) if the number of remaining bits to be punctured is greater than 0, and ending a puncturing process if the number of remaining bits to be punctured is not greater than 0.