%AZhang, T%AAl-Qadi, M%AHui, R%AFumagalli, A%Anull Ed.%D2021%I %K %MOSTI ID: 10292080 %PMedium: X %TExploration of Optical Signal QoT Margin with Intelligent WSS Filtering Penalty Estimator Using Neural Network %XDeployment of 5G requires increased data trans-mission capacity in the metro fiber network. Besides deploying new dark fiber operators are also looking into solutions that improve fiber spectrum utilization by means of high-order modulation formats, flexible grid, and subcarrier multiplexing (SCM) technologies. An important factor that limits fiber spectrum utilization in metro network is the penalty inflicted on the optical signals that are routed by wavelength selective switches (WSS). In this paper, an intelligent WSS filtering penalty estimator is proposed based on neural network. With the achieved accuracy of 0.34 dB of mean absolute error in estimating the optical signal-to-noise ratio penalties caused by WSS filtering, the trained neural network is applied to estimate the fiber throughput gains that can be obtained by optimally selecting the signal symbol rate in a number of use cases. Country unknown/Code not availableOSTI-MSA