Search

We investigate a silicon nitride (Si3N4) horizontal slot waveguide with silicon dioxide (SiO2) cladding, which can achieve a flat and low normal dispersion of −15.56 ps/(nm•km) over a 3270-nm wavelength range from 1170 to 4440 nm, covering a 1.9 octave bandwidth. By launching a 100-fs 90-kW hyperbolic secant input pump pulse, a three-octave supercontinuum (SC) from 504 to 4229 nm can be generated using a 5-mm long waveguide. The generated SC spectrum is highly coherent, as it is mainly expanded by self-phase modulation (SPM) and optical wave breaking (OWB) in all-normal dispersion regime. The degree of coherence can achieve >0.99 from 1250 to 4300 nm. Therefore, the proposed waveguide could potentially serve as a key element in the on-chip broadband source for silicon photonic platform.

In this work, deep learning is used to monitor coherent channel performance with eye diagram measurement. Experiments show that the proposed technique can determine the modulation format, optical signal to noise ratio (OSNR), roll-off factor (ROF), and timing skew of a quadrature amplitude modulation (QAM) transmitter with high accuracy. Trained vanilla convolutional neural network (CNN) and MobileNet can be utilized to jointly monitor above four parameters with >98% prediction accuracy for 32 GBd coherent channels with quadrature phase shift keying (QPSK), 8-QAM or 16-QAM formats, under OSNR from 15 to 40 dB, IQ skew from −15 to 15 ps, ROF from 0.05 to 1. Our proposed deep learning approach outperforms many traditional machine learning methods, such as decision tree, k -nearest neighbor algorithm (KNN), and histogram of oriented gradient (HOG) based support vector machine (SVM). Unlike other optical performance monitoring approaches, the use of eye diagram measurement combined with deep learning could enable joint monitoring of multiple system performance parameters with reduced hardware implementation complexity. Comparing with vanilla CNN, MobileNet has relatively simplified iteration algorithm, thus reduces the requirement on the computational power, while still maintaining high accuracy for classification issues.