NB-LDPC Check Node Decoding with Presorted Syndrome Selection

The patent segments the variable node messages into multiple groups based on their reliability metrics (e.g., sorting messages by magnitude of log-likelihood ratios). This segmentation allows the system to process only the most reliable messages through the full EMS algorithm while using simplified processing for less reliable messages, thereby reducing overall computational complexity while maintaining decoding performance.

The patent applies partial action by selectively applying the full EMS algorithm only to a subset of variable node messages that exceed certain reliability thresholds. For messages below these thresholds, simplified decoding operations are performed. This partial application of the complex algorithm reduces computational burden while preserving the essential error correction capabilities where they are most needed.

The patent performs preliminary sorting and classification of variable node messages based on reliability metrics before the main decoding process. This preliminary action identifies and prioritizes the most reliable messages, allowing the check node processing units to focus computational efforts on these messages first. This pre-processing step enables the system to achieve acceptable decoding accuracy with reduced latency by processing critical messages ahead of less critical ones.

The patent implements periodic processing where check node processing units handle variable node messages in periodic batches based on their reliability rankings. Instead of processing all messages simultaneously or in strict sequence, the system processes messages in periodic cycles prioritized by reliability, which reduces latency for high-priority messages while maintaining overall decoding accuracy through repeated processing cycles.