Optimization of Multi-Pump Raman Amplifier in Ultra-Wideband WDM Systems: A Review

Abstract

The recent increase in interest in Multi Pump Raman Amplifier (MPRA) has been an important field of research in terms of the surge in the demand for broadband and high-capacity optical communications in Wavelength Division Multiplexing (WDM) systems. This review aims to describe a summary of 37 recent articles published on the design, modeling and optimization of Raman amplification in current optical communication systems based on WDM in terms of strengths, limitations and gaps in knowledge. The reviewed articles fall into three major directions: (1) gain and noise figure (NF) optimization strategies, (2) machine learning-based optimization and adaptive control strategies, and (3) quantum and hybrid quantum-classical reinforcement learning systems. The architectural needs of RAs to enable the transmission of WDM systems still offer possibilities to the researchers, and it has the ability of distributing, broadband optical gain with lower noise figures (NF) and flexibility across a variety of spectral bands. Recent explored approaches to machine learning (ML) and evolutionary optimizations to rapid estimation of pump powers will result in solutions enabling the inverse design of pump powers to arbitrary shaped Raman gain profiles, and the production of adaptive pump powers in ultra-wideband (UWB) WDM systems. ML frameworks such as neural networks, particle swarm optimization (PSO), and differential evolution (DE) demonstrate minimal prediction errors with respect to an effective consumption of pump powers to operate the Raman amplifiers. Multi-band programmable amplifier designs, which act as a combination of the hybrid Raman-EDFA designs, enhance optical signal-to-noise ratio (OSNR) and transmission capacity. Recent work has focused on applying quantum reinforcement learning (QRL) methods to improve scalability and convergence of high-dimensional optimization problems. In general, the review points to a shift toward adaptive real-time control, the combination of ISRS-aware optimization, and the hybrid quantum-classical smart schemes of scalable and energy-efficient ultra-wideband optical networks.