A survey on machine learning schemes for fiber nonlinearity mitigation in radio over fiber system

Abstract

Abstract The fifth generation is the most recent generation of communication needed for high data rates. High spectrum availability, low jitter, high reliability, minimal latency, and increased capacity are just a few benefits of 5th generation. Optical fiber supports the 5G network’s backhaul to meet the enhanced capacity and big data rate requirements. When the fiber is used as a propagation medium, a significant number of nonlinearities manifest. These nonlinearity effects in optical fiber communication are among the most detrimental to modern communication systems because it results in various modulator distortions like phase, harmonic, and intermodulation, distortion, adjoining channel noise, and many more undesirable consequences. The primary fiber nonlinear effects in the radio over fiber communication systems are the Kerr nonlinearity and scattering effects, which are caused by alterations in refractive index as a result of the signal. To overcome these limits, several strategies have been put forth. In particular, the nonlinearity during signal modulation, transmission, and detection has attracted a lot of study attention due to the complex physical layer restrictions in RoF systems. One such fascinating potential is machine learning (ML) methods. In this article, we look at recent advances in ML methods for RoF systems, notably those that use ML models to reduce various types of impairments and improve system performance.