Quantization of Recurrent Neural Network for Low-Complexity High-Speed IM/DD System Equalization Based on Neuron Clustering

Neural networks (NNs) are widely employed as effective equalizers in intensity-modulated direct-detection (IM/DD) links due to their excellent ability in dealing with nonlinear channel impairments. However, the complexity concern impedes the real-time application of NN-based receivers. To address th...

Full description

Saved in:
Bibliographic Details
Published in2023 Asia Communications and Photonics Conference/2023 International Photonics and Optoelectronics Meetings (ACP/POEM) pp. 1 - 4
Main Authors Xu, Zhaopeng, Ji, Honglin, Yang, Yu, Qiao, Gang, Wu, Qi, Lu, Weiqi, Liu, Lulu, Wang, Shangcheng, Liang, Junpeng, Li, Jiali, Wei, Jinlong, He, Zhixue, Hu, Weisheng, Shieh, William
Format Conference Proceeding
LanguageEnglish
Published IEEE 04.11.2023
Subjects
Artificial neural networks
complexity
Equalizers
mixed-precision quantization
Neurons
nonlinear equalization
Optical losses
Quantization (signal)
recurrent neural network
Recurrent neural networks
System performance
Online AccessGet full text

Cover

More Information
Summary:Neural networks (NNs) are widely employed as effective equalizers in intensity-modulated direct-detection (IM/DD) links due to their excellent ability in dealing with nonlinear channel impairments. However, the complexity concern impedes the real-time application of NN-based receivers. To address this issue, we propose mixed-precision quantization of recurrent NN (RNN)-based equalizers in a 100-Gb/s 15-km C-band IM/DD system, which saves about 73.3% and 22.4% memory compared with traditional floating-point-based and fixed-precision quantized RNN. A simple and effective neuron clustering approach is proposed to realize mixed-precision quantization of RNN without degrading system performance.
DOI:10.1109/ACP/POEM59049.2023.10369904