Repository logo
 

Research data supporting "Experimental Investigation of Physics- and ML-based QoT Estimation for WDM Systems"


No Thumbnail Available

Type

Dataset

Change log

Authors

Faruk, Md Saifuddin 
Mansour, Mariane 
Laperle, Charles 
O'Sullivan, Maurice 
Savory, Seb J 

Description

The Excel file contains four sheets and each of them associated with the figures 2 to 5 in the paper. The processed data was obtained by analising the captured 400 data each contains seven channel power launch into the fibre and corresponding SNR from the experimental setup described in the paper. Sheet for Fig.2 contains number of training data and corresponding root mean square error (RMSE) for channel 4 which is calculated from 50 test data for three methods based on physics, neural network (NN) and Gaussian process regression (GPR). Sheet for Fig.3 includes the data for RMSE [dB] of all seven channels for the three methods where number of training data for physics based method is 50 and ML-based method is 250. Maximum estimation error data for all seven channel for three methods is in the sheet for Fig.4. Finally, the sheet for Fig.5 contains back-to-back measured SNRs of seven channel and those estimated from physics-based method after 1000 km transmission.

Version

Software / Usage instructions

Matlab was used to process raw data and create plots shown in publication except the neural network results which is implemented in PyTorch.

Keywords

machine learning, optical performance monitoring, quality of transmission

Publisher

Sponsorship
supported by the UK EPSRC for funding via the Programme Grant TRANSNET (EP/R035342/1). This research was performed under the auspices of a Ciena University collaborative research grant
Relationships
Supplements: