Layered LDPC Decoding Schedule for Faster Convergence
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional LDPC decoding methods, such as flooding and layered decoding, require multiple iterations to achieve a target block error rate due to low convergence speed, which hampers throughput in wireless communication systems.
Innovation Solution
A method is introduced that involves identifying variable nodes with high log-likelihood ratios, dividing parity check matrix rows into subsets based on the number of non-zero elements at punctured nodes, and ordering these subsets to generate a schedule for layered LDPC decoding, thereby improving convergence speed.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If conventional flooding or layered decoding methods are used for LDPC decoding, then the decoding process is simple to implement, but the convergence speed is low requiring multiple iterations
Solution Approach 1:
The parity check matrix rows are segmented into multiple subsets based on the number of non-zero elements at punctured variable node indices. This segmentation allows the decoder to process different subsets in different iterations, prioritizing subsets with fewer non-zero elements first. The segmentation strategy accelerates convergence by systematically addressing the most critical subsets earlier in the decoding process, thereby improving convergence speed while maintaining manageable decoding complexity through structured organization.
Solution Approach 2:
The decoding schedule is made dynamic by adaptively determining the processing order of row subsets based on their characteristics (number of non-zero elements at punctured indices). Rather than using a fixed processing order, the system dynamically adjusts which subsets are processed in each iteration based on the current decoding state and subset properties. This dynamic scheduling optimizes convergence speed by flexibly adapting to the specific code structure and channel conditions.
2Productivity
If multiple iterations are performed to achieve target block error rate, then the error rate performance is maintained, but the throughput is reduced
Solution Approach 1:
The method performs preliminary organization of the parity check matrix rows into ordered subsets before the actual decoding iterations begin. This preliminary action pre-calculates and stores the optimal processing order based on the number of non-zero elements at punctured indices, so that during the decoding iterations, the processor can directly follow the pre-determined schedule without additional computational overhead. This preliminary preparation enables faster convergence and higher throughput while maintaining the required block error rate performance.
3Speed
If row subsets are divided and ordered based on non-zero elements at punctured VNs, then the convergence speed is improved, but the scheduling complexity increases
Solution Approach 1:
The method changes the parameter used for scheduling from a uniform or random order to an order based on the number of non-zero elements at punctured variable node indices. By using this specific parameter (count of non-zero elements at punctured positions), the system creates a natural ordering criterion that accelerates convergence. The parameter change transforms the scheduling approach into a more efficient process where subsets with fewer non-zero elements at punctured indices are processed first, improving convergence speed while the parameter itself provides a clear, implementable sorting criterion.
Data Source
AI summary
Decoding low-density parity check (LDPC) codes in a communication system includes identifying a set of indices of variable nodes (VNs) having log-likelihood ratios (LLRs) greater than a threshold, Indices of parity check matrix (PCM) rows are divided into subsets each including a same number of non-zero row elements at indices of punctured VNs. The subsets of the indices of the PCM rows are ordered based on the number of non-zero row elements at the indices of the punctured VNs. A schedule is generated based on the ordered subsets of the indices of the PCM rows. Layered LDPC decoding is performed according to the schedule.