BER and OSNR Based Quality Estimation in Optical Networks using Machine Learning Algorithms

Currently, optical fiber technology has become much more advanced to transmit higher data rates over long-distance transmissions. In this area, the research and development work is growing first. The physical layer impairments are an essential estimation parameter over fiber-optic networks to meet the demand of next-generation technology. A physical layer impairments-aware algorithm is an important factor to improve the quality of optical networks. The optical network includes noise due to amplified spontaneous emission in optical fiber, multiplexer, demultiplexer, and coherent crosstalk. Also, optical switches present BER, and OSNR in WDM/DWDM networks. This paper discusses the estimation of ASE noise for different values of wavelength and bandwidth and a RWA algorithm is based on an OSNR to improve the QoT. Firstly, we discuss the QoT problems in fiber-optic networks, by using different models. Secondly, the best machine learning model, sample data, and the quality of the transmission model are briefly described. The simulation results show that the machine learning-based quality of the transmission estimation model can predict the BER, OSNR, and Q-factor. We demonstrate that the ASE noise power and inversion parameters increase exponentially.