一种基于译码权重参数的端到端信道译码优化方法
By training the transmitter, channel, and receiver of the optical fiber communication system end-to-end, the BP decoding weight parameters are optimized, solving the trade-off between decoding performance and complexity in high-speed optical communication systems, and improving decoding efficiency and receiver sensitivity.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- BEIJING INST OF TECH
- Filing Date
- 2025-02-24
- Publication Date
- 2026-07-17
AI Technical Summary
In high-speed optical communication systems, existing technologies struggle to achieve an effective trade-off between decoding performance and computational complexity, especially in polar code decoding, leading to increased power consumption and latency issues. Therefore, it is necessary to explore adaptive decoding optimization methods to improve the efficiency of channel decoding.
By training the transmitter, channel, and receiver of an optical fiber communication system end-to-end, the weight parameters in BP decoding are optimized, and a weight-sharing BP decoding system based on a polar code deep learning neural network is established, reducing the number of weight parameters in the decoding process and lowering the model complexity.
It achieves the reduction of decoding complexity and storage overhead in high-speed optical communication systems, while improving receiver sensitivity and decoding performance, and adapting to dynamic adjustments of different channel conditions and code parameters.
Smart Images
Figure CN120128268B_ABST