LDPC Decoder Minimum Search Using Flattened One-Hot Comparison
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing LDPC decoding methods, such as the minimum-sum algorithm, are inefficient due to the need for linear or tree-based iterative comparisons, which slow down computation speed in error correction processes.
Innovation Solution
An LDPC decoder and decoding method utilizing a flattened structure for computations, converting input values into one-hot vectors and performing priority decoding to find minimum values across all vectors simultaneously, reducing the need for iterative comparisons.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If linear minimum compare/mux method is used, then implementation is simple, but computation speed is slow
Solution Approach 1:
The patent transforms the comparison problem from a sequential one-dimensional process into a parallel multi-dimensional operation by converting input values into one-hot vectors. This allows simultaneous comparison across multiple dimensions (bits) using vector operations, dramatically increasing computation speed while maintaining implementation feasibility through standardized vector processing units.
Solution Approach 2:
The patent creates multiple copies of comparison logic in the form of one-hot vector representations. Instead of comparing values sequentially, it generates parallel comparison paths where each bit position is compared simultaneously across all input values, achieving speedup through spatial replication of the comparison function.
2Speed
If tree minimum compare/mux method is used, then computation speed is faster, but device complexity increases
Solution Approach 1:
The patent merges multiple comparison operations into a single unified vector operation. By representing all input values as one-hot vectors and performing bitwise OR operations followed by priority decoding, it combines what would require multiple separate comparator stages into one integrated processing unit, reducing device complexity while maintaining high computation speed.
Solution Approach 2:
The patent replaces the mechanical tree-structured comparator network with a more efficient vector-based computational model. Instead of physically arranging comparators in a tree hierarchy, it uses mathematical vector operations (one-hot encoding, bitwise OR, priority decoding) that can be implemented with simpler, more scalable hardware logic.
3Measurement precision
If iterative comparison is used, then all values are compared thoroughly, but time consumption increases
Solution Approach 1:
The patent performs preliminary transformation of input values into one-hot vectors before comparison. This preprocessing step encodes all necessary comparison information in a format that enables parallel processing, so that when the actual comparison is performed, all values are evaluated simultaneously in a single operation rather than through iterative steps.
Solution Approach 2:
The patent maintains continuous useful action by performing comparison operations in parallel without idle iterations. The one-hot vector approach ensures that every input value is actively compared in each processing cycle, eliminating the sequential waiting periods inherent in iterative methods while maintaining complete comparison coverage.
Data Source
AI summary
An LDPC decoder includes a storage unit and one or more computing engine. The storage unit stores an LDPC matrix. The computing engine performs an LDPC decoding algorithm to estimate a decoding result for a codeword based on the LDPC matrix. The computing engine includes a minimum value searching unit to output a minimum value searching result based on a plurality of input values when the LDPC decoding algorithm performs a minimum value searching operation.


