Convolutional Recurrent Machine Learning for OSNR and Launch Power Estimation: A Critical Assessment

Using waveforms from three distinct stages of signal demodulation, we assess performance, computational efficiency, and benefits of using convolutional recurrent neural networks to simultaneously and independently estimate OSNR and launch power within a multi-channel system.