학술논문
Impairment Identification for PAM-4 Transceivers and Links Using Machine Learning
Document Type
Conference
Author
Source
2021 Optical Fiber Communications Conference and Exhibition (OFC) Optical Fiber Communications Conference and Exhibition (OFC), 2021. :1-3 Jun, 2021
Subject
Language
Abstract
We demonstrate simultaneous TDECQ estimation and impairment identification using convolutional neural networks. Robust training with representative impairments yielded identification accuracy of 100% for TDECQ >2.6 dB, when considering limited bandwidth, signal compression and SNR degradation.