Implications of Fiber Nonlinearities on Coherent Optical Fiber Communications

Thumbnail Image
Kashi, Aazar
Fiber Optic Communication , Kerr Fiber Nonlinearities
A primary objective of optical system design is to satisfy the ever-increasing capacity demand for communication services. Fulfilling this target by increasing the cardinality of modulation formats is at the expense of a reduced system tolerance to transmission impairments with the capacity ultimately limited by Kerr fiber nonlinearities. The research presented in this thesis focuses on exploring solutions to fulfill the demand for increased capacity and studying advanced techniques for modelling and characterizing systems, predicting performance, and estimating critical parameters. The thesis investigates the quantification of the impact of fiber nonlinearities on the spectral broadening denoted as the spectral edge power (SEP) (‎Chapter 2). The implications of intra- and inter-channel nonlinearities for 224 Gb/s single-channel and 256 Gb/s 9-channel dual polarization 16-ary quadrature amplitude modulation (DP 16-QAM) systems are quantified in terms of SEP. The experimental results demonstrate that the launch power that minimizes the bit error ratio and SEP coincides. The research utilizes spectral and power efficient multi-dimensional modulation formats, which provide the capability to fine-tune the spectral efficiency (‎Chapter 3). The performance of the four-dimensional set-partitioning (SP) 128-QAM is experimentally investigated. The results demonstrate improvements of 5.3% and 8.2% in the information rates for the 256 Gb/s SP 128-QAM compared to the DP 16-QAM at the same bit rate and for transmission over 2100 km and 2700 km, respectively. The thesis studies the estimation of the nonlinear signal-to-noise ratio (SNRnl) using artificial neural networks (‎Chapter 4). The accuracy of the estimation procedure is verified for a 34.5 Gbaud DP 16-QAM signal with a maximum normalized root-mean-square error of 0.37% for estimating SNRnl in simulation, and less than 0.25 dB deviation from the true SNRnl for estimates obtained with varying fiber length and launch power in experiments. The research explores the identification of the shaping distribution of rate-adaptive transmission systems using the estimates of SNRnl and taking advantage of the unique properties of nonlinear interference noise (‎Chapter 5). The identification results exhibit success rates of higher than 96% and 79% for two groups of five shaping distributions corresponding to bit rate granularities of 25 Gb/s and 12.5 Gb/s, respectively.
External DOI