LDPC Decoder Input Refresh for Error Floor Recovery
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Data processing systems face challenges in error recovery due to the 'error floor' phenomenon, where low density parity check code graphs and decoders fail to converge, leading to high uncorrectable bit error rates and increased latency, especially in flash memory systems.
Innovation Solution
The implementation of a refresh, run, and aggregate decoder recovery method, which involves refreshing the decoder input by altering log likelihood ratio values and re-running decoding iterations, using techniques like log likelihood ratio replacement and scaling, and aggregating decoder outputs to generate new initial inputs, thereby avoiding trapping sets and improving convergence.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the decoder uses standard error correction procedures with LDPC codes, then the decoding process is efficient and fast, but the uncorrectable bit error rate increases due to the error floor phenomenon where the decoder fails to converge
Solution Approach 1:
The patent applies preliminary action by performing a refresh of the decoder input values before re-running the decoding algorithm. This involves initializing the decoder input with new values derived from the likelihood values after the decoder has failed to converge, thereby preparing the decoder in advance for a successful decoding attempt and avoiding the error floor phenomenon
Solution Approach 2:
The patent changes the parameters of the decoder input by altering the initial values used in the decoding algorithm. When the decoder fails to converge, the input values are refreshed using transformed likelihood values, which modifies the decoding trajectory and enables convergence that was not achieved with the original input values
2Reliability
If the decoder fails to converge and recovery features are activated, then the uncorrectable bit error rate can be improved, but the latency increases due to additional decoding iterations
Solution Approach 1:
The refresh operation is performed as a preliminary step before re-running the decoder, preparing optimized input values in advance. This preliminary action reduces the number of iterations needed in subsequent decoding attempts, thereby minimizing the additional latency introduced by recovery features
Solution Approach 2:
The patent skips unnecessary intermediate steps by directly refreshing the decoder input with transformed likelihood values when convergence failure is detected, rather than performing extensive error analysis or multiple incremental adjustments. This rushing through the recovery process minimizes the time penalty while achieving the reliability improvement
Data Source
AI summary
A data processing system includes a likelihood input operable to receive encoded data, a decoder operable to apply a decoding algorithm to likelihood values for the received encoded data and to yield a decoded output, and a decoder input initialization circuit operable to generate new decoder input values based in part on the likelihood values for the received encoded data after the likelihood values for the received encoded data have failed to converge in the decoder.