Optical Receiver LLR Segmentation for Multidimensional Symbols
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The calculation of log likelihood ratios for multidimensional symbols in optical communication systems is computationally intensive due to the need to calculate Euclidean distances for all candidate symbols, which increases the complexity and resource requirements for error correction.
Innovation Solution
An optical receiver is designed with separate calculation units for M-dimensional and N-dimensional log likelihood ratios, where M is less than N, allowing for reduced calculation by using feedback from soft determination error correction decoding to update and combine log likelihood ratios, thereby reducing the number of necessary calculations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If multi-leveling modulation (DP-64QAM or higher) is used to increase transmission capacity, then frequency utilization efficiency is improved, but minimum Euclidean distance between symbols reduces and noise resistance properties decrease
Solution Approach 1:
The patent transitions from traditional 2D signal point arrangement (I-Q plane) to N-dimensional space arrangement. By expanding the design dimension from 2 dimensions to N dimensions (where N≥3), the system can arrange signal points in higher-dimensional space, increasing the minimum Euclidean distance between symbols while maintaining high frequency utilization efficiency. This dimensional expansion allows simultaneous achievement of high transmission capacity and improved noise resistance.
2Reliability
If log likelihood ratio calculation for multidimensional symbols is performed using conventional methods, then soft decision error correction is achieved, but calculation complexity increases due to Euclidean distance computation for all candidate symbols
Solution Approach 1:
The patent segments the N-dimensional symbol calculation into multiple stages: first calculating log likelihood ratios for lower-dimensional symbols (M-dimension where M<N), then progressively combining these results to obtain the final N-dimensional log likelihood ratio. This segmentation avoids the need to calculate Euclidean distances for all candidate N-dimensional symbols simultaneously, significantly reducing calculation complexity while maintaining error correction performance.
Solution Approach 2:
Instead of performing complete Euclidean distance calculations for all candidate N-dimensional symbols, the patent performs partial calculations by first computing log likelihood ratios for subsets of dimensions and then combining these partial results. This partial action approach achieves the necessary error correction performance without the excessive computational burden of complete calculations.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
An optical receiver includes: a first calculation unit that obtains a log likelihood ratio for each M-dimension (M is a natural number), based on a received signal; and a second calculation unit that obtains a log likelihood ratio of an N dimensional symbol (N is a natural number), based on the log likelihood ratio for each M-dimension.