Raman amplification is an all-optical technology that allows for long-distance coverage and will compete with EDFAs in the future optical amplification market. This article describes its main characteristics and design factors.Essentially, Raman amplification involves using commercially available pump lasers to transform optical fiber into a broadband, distributed-gain transmission medium. In other words, the fiber acts as a long optical amplifier, providing gain to the signals propagating within it. Raman amplifiers are typically used in a counter-propagation configuration, meaning the Raman pump wavelengths are introduced from the end of the fiber segment, traveling in the opposite direction to the DWDM signals. This ensures that most of the amplification occurs at the end of the fiber path, where signal levels are lower, while also preventing the power levels at the output of each amplification segment from exceeding the threshold for nonlinear effects (Figure 1).
Like EDFAs, distributed Raman amplification contributes to system noise. However, unlike EDFAs, a different definition is used: the effective noise figure. This concept is equivalent to replacing the combination of distributed Raman amplification and EDFA with a single discrete amplifier and its effective noise figure. It can therefore be seen that Raman amplification reduces the effects of ASE noise generated by EDFAs, improving the optical signal-to-noise ratio (OSNR) of the entire system. For example, effective noise figures of less than 3 dB or even negative values can be obtained, values that are impossible to achieve with EDFAs alone. In general, increasing the Raman gain results in lower effective noise figures, although there is a limit. If the gain is increased excessively, problems arise in the fiber associated with Rayleigh scattering. As the gain increases, more signal power is dispersed inside the fiber, resulting in signals propagating in the opposite direction. These, in turn, can be reflected back and produce "ghost" signals in the receiver that arrive with a certain delay compared to the main signal.
Additionally, Raman amplification is polarization-dependent, such that signals polarized orthogonally to the Raman source do not experience any gain. Therefore, Raman amplifier modules should have a depolarized output to minimize noise. A depolarized source provides the same power on each polarization axis of the fiber, so a WDM channel is amplified by the same amount regardless of its polarization state. In practice, a depolarized source is achieved by polarization-multiplexing two pump lasers, while ensuring precise power balancing between the two sources. Another option is to use a fiber depolarizer. These work by separating polarized light into its two orthogonal components using a polarization-maintaining fiber.
Although the fundamental design principle of a Raman amplifier is simple – each Raman pump wavelength provides gain at a frequency 13.5 THz lower – the interactions between Raman wavelengths and DWDM signals can be complex. This complexity arises from the fact that the same DWDM channels can act as Raman pumps for other signals at higher wavelengths, despite their much lower optical power level. This creates multiple Raman gain regions 13.5 THz below the frequency of each signal (approximately 100 nm in the third window). In conventional systems that only carry traffic in the C-band (1530–1560 nm), these interactions are not particularly significant. However, the situation is different when Raman amplification is used in the presence of both C-band and L-band channels. In this case, the C-band Raman pumps lose a significant amount of power amplifying the L-band Raman pumps. At the same time, the C-band channels provide some amplification to the L-band channels. The result is a certain slope in the optical power of the channels (lower power at shorter wavelengths and higher power at longer wavelengths) that needs to be equalized.
Therefore, C-band Raman pumps require more power than L-band pumps. This must be considered even in the case of a system that only transmits C-band traffic, as future use of L-band would be limited if pump sources lacked the capacity to generate higher power levels. Additionally, the interrelationship between pump power and DWDM signals necessitates adjustments if the power of any pump is modified.
Raman Effect:
When an optical pump signal encounters an optical fiber, it experiences a degree of spatial scattering caused by molecular vibrations (or phonons). This scattered light is shifted in frequency by an amount equal to the difference between the pump frequency and the molecular vibration frequency of the crystal. The shifted signal is known as the Stokes field, while the nonlinear effect is called Raman scattering or scattering. This Stokes field can be remixed with the optical signal to provide additional frequencies. In turn, these signals drive the crystal vibrations. The end result is a stronger Stokes field, and the process is called stimulated Raman scattering.
The Stokes shift determines the frequency (relative to the original signal) at which Raman gain occurs. This frequency depends on the material, and in the case of silica fibers, it is approximately 13.5 THz. This means that a typical pump at 1440 nm will produce gain around 1550 nm. Due to the molecular structure of the crystal, there are several molecular vibration frequencies, so the gain region has a spectral width of approximately 30 nm. Thus, it is possible to achieve gain for any wavelength in a standard fiber simply by appropriately selecting the pumping wavelength(s). Figure 2 graphically illustrates this design aspect.
The nonlinear scattering that leads to Raman amplification is weak in silica optical fibers, so long lengths (several kilometers) are needed to obtain a reasonable gain value. However, other types of fibers, such as dispersion-shifted fibers, can also be used, where the nonlinear effects are more pronounced.
Pump Lasers:
The design of a Raman amplifier module is somewhat simpler than that of a broadband EDFA. Once the wavelengths and pump powers have been chosen, it becomes a question of how to design a unit capable of generating these outputs and multiplexing them within a fiber system. The module basically consists of a series of laser sources, a multiplexing scheme (wavelength and polarization), an optical monitoring system, and a fiber multiplexer (Figure 3).
Pump laser sources are the main driver of the Raman amplifier market. Pump lasers with high output powers in the 14xx nm wavelength range are needed for C-band and L-band signal amplification. Several manufacturers offer high-efficiency pump lasers with powers between 300 and 500 mW (Figure 4). Lasers with power outputs exceeding 1 W at multiple wavelengths and over 700 mW in fiber optics have also been demonstrated. Additionally, some companies market hybrid-integrated pump modules that deliver over 400 mW. For example, two polarization-multiplexed 400 mW pump laser chips can provide 720 mW at a single wavelength with a depolarized output beam. Finally, modules with multiple wavelength-multiplexed pump lasers are also available. Eye safety is a major concern associated with Raman amplifiers, as they are typically Class 4 laser devices (output powers exceeding 500 mW).
To mitigate risks in the event of a fiber break, a reflection monitor is installed inside the units.
Additionally, high stability and wavelength accuracy of these lasers are essential, for which Bragg grating-based locking techniques are typically employed. Since temperature significantly impacts performance, the laser chip temperature is controlled by thermoelectric coolers (TECs), which can dissipate power exceeding 10 W.
Ultimately, given that the performance and reliability of Raman amplifiers depend heavily on the pump lasers, and these constitute a significant portion of the cost, energy consumption, and space requirements, new generations of pump lasers will be crucial for the success of this type of amplifier.
Author:
Francisco Ramos Pascual. PhD in Telecommunications Engineering
Full Professor at the Polytechnic University of Valencia
