LDPC Decoder Message Scaling Under Saturation Constraints
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current LDPC decoding methods face challenges in achieving an ideal compromise between decoding performance, data rate, complexity of implementation, and energy consumption, particularly in high-speed communication systems like space communications, where error correction power, throughput, and hardware complexity are critical.
Innovation Solution
The method involves executing iterations of LDPC decoding with on-the-fly scaling of parity check messages and a posteriori estimation variables when saturation criteria are met, which includes scaling to the nearest integer greater than half the value's absolute, and adjusting thresholds and iteration counts based on coding rates, to reduce error rate floors and memory footprint.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If fixed-point representation is used to reduce hardware complexity, then device complexity is reduced, but measurement precision deteriorates due to quantization floor
Solution Approach 1:
The patent implements dynamic scaling of data values during the decoding process. When saturation is detected in fixed-point calculations, the system automatically adjusts the scaling factor to prevent overflow and maintain precision. This dynamic adaptation allows the system to maintain high decoding precision while using fixed-point representation, resolving the contradiction between hardware simplicity and measurement accuracy.
2Reliability
If more iterations are performed to improve decoding performance, then reliability is improved, but productivity deteriorates due to increased latency
Solution Approach 1:
The patent employs a saturation detection mechanism that monitors the decoding process and provides feedback on calculation overflow conditions. When saturation is detected, the system triggers scaling operations and adjusts iteration behavior. This feedback loop enables the system to achieve reliable error correction by adapting the number of iterations based on actual decoding progress and saturation conditions, rather than blindly increasing iteration count.
Solution Approach 2:
The system dynamically changes parameters during decoding, including scaling factors and iteration continuation decisions. When saturation is detected, the scaling parameter is adjusted to prevent overflow, and the decision to continue iterations is made based on whether saturation has been resolved. This parameter adaptation allows the system to balance reliability and throughput by avoiding unnecessary iterations when saturation indicates convergence.
3Measurement precision
If saturation prevention scaling is applied to maintain precision, then measurement precision is improved, but device complexity worsens due to additional scaling operations
Solution Approach 1:
The patent implements a self-service saturation prevention mechanism where the system automatically detects saturation conditions and applies scaling operations without external intervention. The saturation detection logic is integrated into the existing decoding architecture, and scaling operations are triggered automatically when overflow conditions are detected. This self-service approach maintains precision while minimizing additional complexity by reusing existing hardware resources.
Data Source
Figure 1~2
Figure 3
Figure 4
AI summary
The invention relates to an LDPC decoding method (100). The method involves performing iterations until a stop criterion is satisfied. Each iteration involves: - calculating (111) variable messages (αn,m), - calculating (112) parity check messages (βm,n), - calculating a posteriori estimation variables (γn), the parity check messages (βm,n) and the a posteriori estimation variables (γn) being saturated at a predetermined maximum value. At the end of an iteration, when the number of saturations reaches a specified threshold, the method involves at least a first scaling (152) of the parity check messages (βm,n) and the a posteriori estimation variables (γn). Scaling corresponds to assigning, to a value, an integer which has the same sign and whose absolute value is the nearest integer greater than the absolute value of the value divided by two.