While network architects try to counter the effects of non-linearities, component manufacturers can use these properties to their advantage.
Stuart Woods
Southampton Photonics Inc.
Specialty optical fibers have fundamentally changed optical telecommunications, the most notable example being the erbium-doped fiber amplifier (EDFA) based on an erbium-doped fiber. However, many other possibilities for application-specific component-level specialty fibers exist, from the well-known polarization-maintaining fiber to others still in the research phase -- for example, non-linear fibers for wavelength conversion applications, high numerical aperture (NA) active fibers for sources, metal-electroded fibers for poled-fiber modulators, spun fibers to reduce birefringence and spun hibi fibers for optical current sensors, polarizing fibers, helical-core fibers, numerous rare earth doped fibers, and ring-doped fibers for high-power fiber lasers.
These fibers are enabling a diverse new generation of components and subsystems with radical new economics for tomorrow's telecommunications networks. Non-linear fibers appear to be most promising in the near term to enable this new economic age of telecommunications components.
Non-linear effects
To best understand how we control non-linear effects in a fiber for component applications it is important to understand how these effects have plagued us in telecommunications. Non-linear effects are the ultimate limits to long-haul transmission performance.
As systems operate over increasing distances and channel counts, non-linear effects become more critical. In contrast, other effects encountered in long-haul transmission have been overcome -- attenuation with amplification and dispersion with compensation, for example. But non-linearities just accumulate. Remember that as we increase channel count or distance we are increasing the power in the system. Non-linearity effects increase with the square of the optical power; therefore, the accumulation gets worse as networks expand.
Non-linearities can affect both single-channel operation as well as densely populated transmissions such as DWDM. In the case of single-channel transmission, the pulses go into the fiber and come out smeared, similar to dispersion effects. In multiple-channel transmission, there is both a single-channel effect and a new effect of transference of energy between the different wavelengths. In this transference, energy is shared, or in other words, energy leaks from one wavelength into another as cross talk.
Furthermore, this sharing forms new intermodulation or mixing wavelengths as products of this interaction. This interaction is nothing new to the optics or harmonics, as these are the same kind of intermodulation products associated with harmonic distortion in a stereo system. This effect in optics is called four-wave mixing and is due to the generation of a "non-linear" index of refraction in the fiber (Figure 1).
It is the non-linear index of refraction that is of most interest in the fabrication of a component fiber. In general, the index of refraction is the ratio between the speed of light in a vacuum and the speed of light in the material -- in this case, the glass.
But the index of refraction is not constant; it is a function of the intensity of the light in the fiber from the multiple pulses. Therefore, as the intensity of the optical pulses increases so does the change or modulation of the index of refraction. By reducing the area of interaction in a fiber to force high intensity in a small area, we can effectively force the modulation of the index of refraction and therefore the generation of our new wavelength in a very short distance. In fiber, the measure of the area of interaction is the mode field. The lower the mode field, the more interaction we can expect.
For four-wave mixing to be seen, there is another element that must be present. As we have said, the two wavelengths must travel down the fiber next to each other for a long distance to exchange energy, but they must also travel at the same speed. Dispersion causes different wavelengths to propagate at different speeds; therefore, the more dispersion, the more the two wavelengths walk off from each other and the less interaction through four-wave mixing is generated. Therefore, in designing a non-linear fiber, we look to hold the dispersion as low as possible to maintain the speed of the pulse and maximize this interaction.
As discussed above, by increasing the intensity of light in a fiber and reducing the dispersion, we can increase the likelihood of four-wave mixing or intermodulation between wavelengths in our fiber. In normal long-haul systems with typical non-zero dispersion-shifted fiber and dispersion control, it takes up to well over 500 km for four-wave mixing to start to have a system penalty. By designing a fiber specifically for this interaction, this interaction length can be reduced to 50 cm!
Benefits of non-linear fibers
With a specialty fiber that can generate four-wave mixing intermodulation products within a 50-cm span, there are many possible applications. The most likely is a low-cost component for data copying or wavelength conversion.
If we input two wavelengths into our non-linear specialty fiber, we will get not only the input wavelengths but also their respective four-wave mixing intermodulation products. All the new intermodulation wavelengths can be calculated from inputs into the specialty fiber; therefore, we can predict where they will fall. Because we can predict where they fall, we can then filter out unwanted wavelengths and only allow the wavelength we want to pass to exit our new specialty-fiber-based component. Filtering is best achieved with a fiber Bragg grating (Figure 2).
We can also control the wavelengths that enter our component through filtering before the component or by the introduction of new wavelengths at the component -- all technology that is common in today's networks. By controlling the input and output of our non-linear fiber, we now have a component that is not based on exotic materials with mechanical switches, switching at microseconds, supported by vast amounts of network management, but a cost-disruptive component ideal for nodes on a network where data must be switched or converted to a new wavelength simply based on fiber.
In conclusion, non-linear fibers appear to be a viable marriage of the natural laws of physics and our need to expand and control our telecommunications systems with cost-disruptive solutions. Like erbium-doped fibers, this is a marriage that has great promise in the near future.
Stuart Woods is business development director at Southampton Photonics Inc. (Southampton, UK).