LAUSR

Abstract In this paper, we propose an in-band optical signal-to-noise ratio (OSNR) monitoring method based on support vector machine (SVM). This method is tailored for the analysis of polarization-division multiplexing (PDM) m-ary quadrature amplitude modulation (m-QAM) signals. OSNR-related features are extracted from the numerical characteristics of Stokes parameters, while modulation-order related features are extracted from the eye diagram of s 1 , which is one of the Stokes parameters. The feasibility of this approach is verified through numerical simulations. The simulation results show that the root mean square error (RMSE) of OSNR monitoring for 28 GBaud PDM-mQAM (m = 4, 8, 16, 32) is within 1 dB over an OSNR range of 15–35 dB at back-to-back transmission. Furthermore, the RMSE of OSNR monitoring keeps within 1 dB under the interference of polarization impairments and up to 4800-ps/nm chromatic dispersion (CD). Since the proposed OSNR monitoring method does not require either additional local laser or sophisticated DSP for signal recovery, it features relatively low cost and complexity with high performances, promising for potential applications in the flexible and elastic optical networks.