Selective LDPC Decoding with Partial Maximum Likelihood Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
High rate low density parity check codes in data processing systems suffer from a severe error floor due to short cycles, which degrade the performance of belief propagation decoder circuits, while maximum likelihood decoders are too complex for practical implementation.
Innovation Solution
A combination data decoder circuit that includes a low complexity decoder, such as a belief propagation decoder, and a partial maximum likelihood decoder, which selectively applies a maximum likelihood algorithm to correct errors when a potential trapping set is detected, reducing overall complexity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a belief propagation decoder is used, then the device complexity is reduced, but the reliability deteriorates due to severe error floor caused by short cycles
Solution Approach 1:
The decoder is segmented into two functional parts: a belief propagation decoder for initial decoding and a maximum likelihood decoder for targeted error correction. This segmentation allows each component to operate in its optimal regime, with the simpler BP decoder handling most cases and the more complex ML decoder only when needed, thus resolving the contradiction between overall complexity and reliability.
Solution Approach 2:
Instead of applying the computationally intensive maximum likelihood decoding to all data, the system applies it partially only to subsets of decoded output where errors are detected. This partial action approach maintains high reliability for critical error cases while avoiding the excessive complexity of full ML decoding for all data.
2Reliability
If a maximum likelihood decoder is used, then the reliability is improved, but the device complexity becomes too high for practical implementation
Solution Approach 1:
The system segments the decoding task so that maximum likelihood decoding is applied only to specific subsets of data where errors are detected, rather than to all decoded output. This segmentation dramatically reduces the overall computational complexity while maintaining high reliability for error-prone cases.
Solution Approach 2:
The belief propagation decoder acts as an intermediary that first processes the decoded output and identifies error locations. This intermediary step enables the subsequent maximum likelihood decoding to focus only on problematic areas, reducing its complexity burden while maintaining its high reliability benefits.
3Reliability
If iterative decoding is performed multiple times, then the reliability is improved, but the loss of time increases due to excessive iterations
Solution Approach 1:
The system performs iterative maximum likelihood decoding only partially, applying it only when and where errors are detected by the belief propagation decoder. This avoids excessive iterations on already correct data, significantly reducing time loss while maintaining reliability improvement where needed.
Solution Approach 2:
The system uses feedback from the belief propagation decoder to control whether iterative maximum likelihood decoding is applied. The BP decoder's output serves as feedback that triggers or prevents further iterative processing, optimizing the balance between reliability improvement and time consumption.
Data Source
AI summary
Various embodiments of the present invention provide systems and methods for data processing. For example, a data processing system is disclosed that includes a combination data decoder circuit. The combination data decoder circuit includes a first decoder circuit and a second decoder circuit. The first decoder circuit is operable to apply a first data decode algorithm to a decoder input to yield a decoded output. The second decoder circuit is operable to apply a second data decode algorithm to a subset of the decoded output to modify at least one element of the decoded output to yield a modified decoded output.