Multi–Band Programmable Gain Raman Amplifier

Abstract

Optical communication systems, operating in C-band, are reaching their theoretically achievable capacity limits. An attractive and economically viable solution to satisfy the future data rate demands is to employ the transmission across the full low-loss spectrum encompassing O, E, S, C and L band of the single mode fibers (SMF). Utilizing all five bands offers a bandwidth of up to ∼53.5 THz (365 nm) with loss below 0.4 dB/km. A key component in realizing multi-band optical communication systems is the optical amplifier. Apart from having an ultra-wide gain profile, the ability of providing arbitrary gain profiles, in a controlled way, will become an essential feature. The latter will allow for signal power spectrum shaping which has a broad range of applications such as the maximization of the achievable information rate × distance product, the elimination of static and lossy gain flattening filters (GFF) enabling a power efficient system design, and the gain equalization of optical frequency combs. In this paper, we experimentally demonstrate a multiband (S+C+L) programmable gain optical amplifier using only Raman effects and machine learning. The amplifier achieves >1000 programmable gain profiles within the range from 3.5 to 30 dB, in an ultra-fast way and a very low maximum error of 1.6 • 10 −2 dB/THz over an ultrawide bandwidth of 17.6-THz (140.7-nm). Index Terms-optical communications, multi-band systems, optical amplifiers, machine learning, neural networks. I. INTRODUCTION O VER the past two decades, a great evolution of optical communication systems, in terms of spectral efficiency×distance product, has been enabled by the advances in digital coherent detection. So far, most of the efforts, on reaching the capacity of the nonlinear fiber-optic channel, have been focusing on the C-band

Analysis of Available Components and Performance Estimation of Optical Multi-Band Systems

“…An advantage of Raman amplifiers is that they can create arbitrary gain profiles at will, using machine learning methods, which can generate the desirable output power even in a channel basis. In this way, the gain can be tailored to each band to ensure an optimal physical layer performance in all bands of an OMB system concurrently [16]. Other significant advantages of the Raman-type amplifiers are the distributed amplification and their high output power (about 30 dBm), although such an output power may not be always desirable due to non-linear phenomena and due to safety reasons.…”

Simultaneous gain profile design and noise figure prediction for Raman amplifiers using machine learning

“…Providing an arbitrary gain profile, in a controlled way, requires selecting the appropriate pump powers and frequencies. We have introduced and experimentally validated a machine learning (ML) framework for determining the corresponding pump powers and frequencies, given a target gain profile [5][6][7]. This proved to be a versatile tool, able to rapidly provide highlyaccurate gain designs for different amplifier schemes such as C [5], C+L [6], and S+C+L-band amplifiers [7], for discrete and distributed configurations.…”

“…We have introduced and experimentally validated a machine learning (ML) framework for determining the corresponding pump powers and frequencies, given a target gain profile [5][6][7]. This proved to be a versatile tool, able to rapidly provide highlyaccurate gain designs for different amplifier schemes such as C [5], C+L [6], and S+C+L-band amplifiers [7], for discrete and distributed configurations. Similar methods were also used to design the RA in hybrid approaches with EDFA [4] or semiconductor optical amplifier (SOA) [8], and few-mode RA [9].…”

“…So far, the focus has only been on the design of target gain profiles without consider the noise properties [4][5][6][7][8][9]. However, different pump configurations in terms of frequency and power can affect both the signal and the noise power spectral densities during the Raman amplification.…”

Inverse design of a Raman amplifier in frequency and distance domains using convolutional neural networks

“…EDFAs are more power efficient while DRAs provide low NF. Moreover, DRAs' power profile can be adjusted easily by changing the power and wavelength of the pumps [1, 2], which makes them more attractive for wideband wavelength division multiplexed (WDM) scenarios [3].…”