LDPC Decoder Message Scaling for Saturation-Limited Precision

The patent implements dynamic scaling of parity check messages based on the number of iterations performed. The scaling factor is adjusted adaptively during the decoding process, transitioning from no scaling in early iterations to active scaling when saturation is detected. This dynamic approach allows the system to maintain high decoding precision when needed while avoiding the overhead of continuous scaling operations, thus resolving the contradiction between hardware complexity and decoding precision.

The patent changes the parameter of message scaling dynamically based on iteration count and saturation detection. Instead of using a fixed scaling factor or always applying scaling, the system adjusts the scaling parameter adaptively - using identity scaling (no operation) when not needed and applying appropriate scaling factors when saturation occurs. This parameter change strategy maintains precision while minimizing hardware complexity.

The patent applies message scaling partially rather than continuously. Scaling is triggered only when saturation conditions are detected (when the number of saturated messages exceeds a threshold), and only for specific iterations (typically after a certain number of iterations have passed). This partial application of scaling reduces the processing complexity and computational overhead while still achieving the reliability benefits of error floor reduction when actually needed.

The system monitors its own state (saturation of messages) and automatically activates scaling when needed without external control. The saturation counter and iteration tracker enable the decoder to self-regulate the application of scaling operations, applying them only when the system detects that saturation is occurring and precision is degrading. This self-service mechanism improves reliability while avoiding unnecessary processing complexity.