Neural Min-Sum LDPC Decoder With Self-Corrected Offset Tuning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current decoding methods for Low-Density Parity-Check (LDPC) codes in wireless communication systems face challenges in achieving efficient error correction with minimal computational complexity, particularly in finding optimal offsets for offset min-sum decoding.
Innovation Solution
An electronic device and method that employs a neural network for real-time optimization of decoding factors, incorporating a self-correction technique to adjust information deletion and dropout rates, thereby enhancing the decoding performance of LDPC codes through iterative decoding and hysteresis-based operations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If sum-product decoding is used for LDPC codes, then decoding performance is improved, but computational complexity increases significantly
Solution Approach 1:
The patent replaces the complex sum-product decoding algorithm with a simpler min-sum decoding algorithm that uses approximate computations. Instead of performing precise arithmetic operations required by sum-product decoding, the min-sum algorithm uses simplified logarithmic domain operations that are computationally cheaper and can be implemented with lower complexity hardware or software structures.
2Reliability
If offset min-sum decoding is used to approach sum-product performance, then decoding performance is improved, but the effort to find optimal offset for each case increases
Solution Approach 1:
The patent introduces a learnable offset parameter that is optimized through neural network training rather than manual optimization for each code case. The offset parameter is adjusted dynamically based on the input data characteristics, allowing the system to adapt to different channel conditions and code parameters without requiring separate manual optimization processes. This transforms the static offset selection into a dynamic, data-driven parameter adjustment mechanism.
Solution Approach 2:
The patent implements a self-correction mechanism where the neural network automatically adjusts the offset parameter based on decoding performance feedback. The system monitors decoding outcomes and uses this information to refine the offset parameter through backpropagation and gradient descent, enabling the system to self-optimize without external intervention for each new code configuration or channel condition.
3Reliability
If neural network is used for real-time optimization of decoding factors, then decoding performance is improved, but computational complexity increases
Solution Approach 1:
The patent performs neural network training offline beforehand to generate optimized decoding factors and offset parameters for various channel conditions and code configurations. These pre-computed factors are stored and directly applied during real-time decoding operations, eliminating the need for complex neural network computations during actual decoding. This shifts the computational burden from the real-time decoding phase to an offline training phase.
Solution Approach 2:
The patent implements a hybrid approach where simple min-sum decoding operations are used for the main decoding process, while neural network-based offset adjustment is applied selectively based on channel conditions and decoding performance. The system dynamically switches between fixed offset values and learned offset adjustments, applying neural network computations only when beneficial, thereby balancing performance improvement with computational complexity management.
Data Source
AI summary
An electronic device and an operating method of an electronic device are provided. The operating method includes configuring a self-correction condition for adjusting an information deletion and dropout rate, performing iterative decoding on the received information using decoding factors and a self-correction technique, determining whether decoding of the codeword succeeds or fails, based on a result of the decoding, storing a received signal and the codeword which are successfully decoded, based on a determination result, and optimizing the decoding factors, based on the stored received signal and codeword.