Impact of filtering in optical coherent systems with finite-length equalizers: modeling and experimental verification

Modern ultra-high bit rate coherent transmission systems frequently operate under conditions where stringent optical and/or electrical filtering becomes a significant source of performance penalty. This is particularly evident when the end-to-end equivalent analog optoelectronic bandwidth falls below the transmitted signal symbol rate. Building upon our prior research, we introduce a novel semi-analytical model designed to quantify the performance degradation induced by filtering and colored-noise in coherent systems. Crucially, the model incorporates the effects of finite-length adaptive equalization at the receiver, thereby more accurately capturing the characteristics of realistic transceiver implementations. This developed tool offers not only a valuable and computationally efficient predictor of system performance, but also serves as a relevant resource for the design of networks and flexible transceivers. Furthermore, it facilitates rapid estimation of the impact of network variations, such as re-routing (which influences optical bandwidth) and power-saving mode operation (which affects idle equalizer taps), thus contributing to the development of real-time performance estimators. In this manuscript, we present, for the first time to the best of our knowledge, a comprehensive experimental verification of the proposed semi-analytical model. We consider the combined influence of Received Optical Power (ROP), Optical Signal-to-Noise-Ratio (OSNR), transceiver noise, and filtering on system performance. Additionally, we validate the model against extensive Monte Carlo numerical simulations for a distributed noise scenario. Our results demonstrate high model accuracy, achieving an estimation error below 0.15 dB compared to statistical numerical simulations, and, with respect to experimental measurements, an error below 0.35 dB in 90% of the cases, and below 0.6 dB in all the cases.