Fiber optic communications systems

have increased exponentially in capacity and distance over the last decade. Commercial implementation (red) of new systems has followed the first research results (blue) by about four years. Figure 2

studies have been carried out on non-linearities such as stimulated Brillouin scattering, stimulated Raman scattering, self-phase modulation, cross-phase modulation and four-photon mixing. Even at power levels as low as a few milliwatts one must now take these nonlinearities into account in designing a system. Despite these limitations, testbeds have demonstrated excellent performance at 5 Gb/sec over 9000 km.

14 Physicists have, however, yet another trump card to play. It turns out that a solution to Maxwell's equation in a lossless, single-mode optical fiber including the nonlinear term and chromatic dispersion is nondispersive in both the time domain and the frequency domain. The stable solution of this type for the optical pulse, called a "soliton," is u{z,t) = sech(f) exp(iz/2), where u{z,f) is the envelope of the pulse, z is the distance of propagation, and t is the elapsed time (both suitably normalized). If one launches such a pulse, with a correctly chosen width-to-peak-power relationship, into an ideal lossless fiber, it will propagate without change over arbitrarily long distances. Physically a pulse of this kind will maintain its shape because the chromatic dispersion and nonlinearity effects cancel each other out. Linn Mollenauer, Rogers Stolen and James Gordon at AT&T Bell Labs demonstrated soliton propagation in fibers experimentally in 1980.

15 For several years research on optical soliton propagation in fibers continued with little attention from optical systems designers, but the invention of the optical fiber amplifier has made long-distance soliton propagation a practical reality. In recent experiments solitons have retained their precise pulse shapes over thousands of kilometers. Furthermore, experiments with two wavelength channels showed that solitons of different wavelengths can pass through each other without changing shape. In these experiments the limits on distance and bit rate (per channel) were set by amplified spontaneous emission from the erbium amplifiers. The amplified spontaneous emission frequency modulates the signal frequencies via the nonlinearity, resulting in a distribution of pulse arrival times ("jitter"). This is known as the Gordon-Haus effect, after Gordon and Hermann Haus (MIT). This jitter can be reduced by using a guiding frequency filter, which guides the pulse in the frequency domain. The improvement is limited, however, by the need for more amplification (to compensate for the loss caused by the insertion of the filter), which in turn leads to more amplified spontaneous emission.

16 A major advance in 1992 overcame this limit and allowed error-free propagation of solitons over a distance of 20 000 km at a bit rate of 10 Gb/sec and over 13 000 km at 20 Gb/sec using two wavelength channels each at 10 Gb/sec. Mollenauer and coworkers achieved this breakthrough by using a sliding frequency filter in which the guiding filter frequency is slightly shifted after each amplification step. This concept creates a transmission line that is transparent to solitons, which can adjust to the frequency shift, but opaque to amplified spontaneous emission noise, which cannot adjust. Since this discovery has overcome the previous limitations of the Gordon-Haus effect, more research is now necessary to establish the new fundamental limit for soliton transmission! Solitons, with all of their advantages, are expected to find their way into commercial systems before the end of the decade.

Fiber lasers

17 Until recently the term "optical fiber components" implied fused-fiber couplers, splitters and multiplexers, fiber Fabry-Perot structures, polarization controllers and so on. Most of these devices are commercially available. A new family of passive fiber devices was made possible by the discovery that ultraviolet light absorbed in the core of germanium-doped silica fiber (conventional transmission fiber) changes the refractive index of the glass. We do not yet understand the detailed microscopic mechanisms, although germanium defect centers play an essential role. The effect has much practical significance, because the induced index changes are large (Δn≥10^-2) and substantially permanent. By illuminating the optical fiber from the side with interfering excimer laser beams at 257 nm, one can record phase gratings directly into the core of the fiber. ^u One can use such a grating to modify the transmission of light propagating in the fiber at a well-defined wavelength.

18 A number of interesting devices based on grating filters and grating reflectors have now been fabricated. Of particular interest for communications is the fiber laser, which consists of a short length of erbium-doped amplifier fiber containing two reflection gratings to define the resonator. Researchers have demonstrated stable, single-mode lasers about 2 cm long operating near 1.53 µ т. Because the gratings transmit the pump laser wavelength (0.98 µ т or 1.48 µ т), the same semiconductor laser can pump the fiber laser and atandem fiber amplifier for high power output. These lasers cannot be directly modulated like semiconductor source lasers and thus require external LiNbO₃ modulators. However, the precision with which the wave length can be set and their reduced temperature sensitivity make these lasers attractive candidates for communications, if they are found to satisfy long-tern reliability requirements.

19 In a broader context the field of fiber lasers is now undergoing rapid growth. Recent achievements include

> fiber lasers emitting up to 5 watts of power at l.06 μm pumped with GaAs diode laser arrays.

> mode-locked fiber ring lasers at 1.55 μ m for soliton sources

> upconversion lasers (based on fluoride fibers) that efficiently convert infrared lasers to the red, blue and green regions of the visible spectrum for optical storage
and display.

All of these fiber lasers are based on the rare earth ion spectroscopy that was studied extensively some two to three decades ago. Because of the simplicity of such devices we can expect a rapid evolution of this field.