Photonics & the Future of Fibre
© Mercury Communications Ltd - March 1993
Issue #3 of Technology Watch looked at basic fibre technologies. This issue examines three technologies that, in combination, are capable of transporting and switching digital data over intercontinental distances at rates in excess of 10Gbit/s: optical amplifiers, solitons, and soliton switching. The bulk of the research & development on these advanced fibre technologies has been undertaken by manufacturers heavily involved in submarine cables, who need to significantly increase data rates, decrease costs and reduce the number of repeaters needed. Laying and maintaining submarine cables is expensive, and any reduction in capital and operational expenses makes them more attractive than other global data transmission media.
Advanced fibre technologies could reduce long-haul transmission costs by factors of between ten and one hundred so their potential impact on the telecommunications industry should not be underestimated!
Today's Submarine Cables
The last analogue submarine cable systems were deployed in the mid 1980s and low-loss digital optical fibre cable has come to dominate transoceanic transmission over the last decade. On the highly used transatlantic route, which tends to lead technological developments, system capacities have moved quickly from the 280Mbit/s of TAT-8 (1988) through 420Mbit/s of C&W's PTAT-1 (1989), to the regenerative 560Mbit/s of TAT-9 deployed in 1991.
All current transoceanic systems use amplifiers called repeaters to overcome fibre attenuation and signal distortion (dispersion). These are placed at regular distances throughout the cable span. Early systems operated at wavelenghts of 1300nm with repeater spacings of 50km. With the advent of 1500 nm technology repeater spans have increased to more than 100km.
Regenerative repeaters are complex opto-electronic devices. They convert (demodulate) the optical signal traffic to electronic signals, filter and amplify them, and convert back to optical signals before relaunching them on the next span of cable. Repeaters need to be designed for a specific data-rate, so upgrading them to higher data-rates at a later date is well nigh impossible.
Figure 3 - Submarine Cable Application Segments
Consequently, a great deal of effort has gone into finding a way of amplifying signals at the optical level wihout converting signals back to electrical signals.
It is not widely known beyond submarine transmission specialists that optical amplifiers are already a reality. Companies like STC Submarine Systems in the UK, AT&T in the USA, and Kokusai Denshin Denwa (KDD) in Japan are now moving out of the system trial phase and moving rapidly towards full scale production.
Optical amplifiers are devices that can be used instead of repeaters in long-haul fibre links. They operate entirely in the optical domain and do not employ any translation of data back to electrical baseband signals. Although there are optical amplifiers based on semiconductor lasers, the principal technolgy used on long-haul links is optical fibre doped with the rare-earth element Erbium, Er (doping is the introduction of minute amounts of an element). These Erbium-doped fibre amplifiers (EDFAs) offer gains of up to 30dB (1000 times) and are essentially transparent to the bit-rate of the data transmitted.
Figure 4 shows the basic arrangement of an EDFA. A high-powered 'pump laser' operating at a wavelength of 1480 nm is mixed with the 1550 nm optical traffic signal and the combined signal is fed into the special amplifying Er-doped fibre. The isolator allows the amplified 1550 nm signal to leave the amplifier but prevents the 1480 nm pump signal from leaving, in effect forcing it to recirculate around the Erbium-doped fibre. Often, the amplifier will also contain a bandpass filter to prevent the build-up of noise which would degrade the signal-to-noise ratio (SNR) of the system.
Figure 4 - An Erbium Doped Optical amplifier
The supervisory circuitry continually monitors the performance of the amplifier and can detect malfunctions or broken cables. When mounted in an undersea cable, each amplifier's supervisory circuit can be remotely accessed to check performance. Usual configurations contain duplicated amplifiers and run the pump-laser at around half-power to lengthen laser life.
The Physics of Optical Amps
The result of optical amplification is simple to describe: the high optical energy of the 1480 nm pump laser is converted by the Er-doped fibre into optical energy at the traffic wavelength of 1550 nm. The result of this is that the output power of the optical-traffic signal is much higher than the input-traffic signal. How it achieves this is less easily understood!
The pump laser source is mixed with the signal traffic using an optical coupler. The combined signals are fed into the Er-doped fibre where the photons from the 1480 nm pump laser signal are absorbed by the rare-earth element Erbium incorporated in the fibre in minute quantities.
The Erbium ions are raised to a higher level of energy as a result (Figure 5). However, the ions that reach this higher level rapidly decay to a lower metastable state (metastable means unstable). The energy difference between the higher level and the metastable level is exactly equal to the wavelength of the 1550 nm signal (that is why Erbium is used). The arrival of a 1550 nm signal photon triggers the release of an Erbium ion temporarily held in the metastable state which then drops down the zero-energy, or ground state where it originated. In doing so, the ion releases energy in the form of a photon at the same wavelength and phase as the 1550 nm signal frequency photon. This mechanism provides the amplification function.
Benefits of Optical Amplification
Optical amplifiers are seen by many as a panacea for many transmission ills, in that they supposedly allow operators to increase bit-rates as and when dictated by traffic requirements. Current technology does not live up to that ideal, unfortunately. For example, a 2.5Gbit/s link requires optical amplifiers to be spaced at 60 - 80km. If it was decided at a later date to upgrade this to 5Gbit/s then amplifiers will need to be spaced at a nominal 40 - 50km. Furthermore, bandwidths are currently limited to around 30nm per amplifier, which reduces considerably as more amplifiers are placed in series. Therefore, a 3000km submarine cable might only achieve a bandwidth of only 3nm allowing only one light carrier to be used. Of course, over time, these limitations will be overcome. Even with this limitation optical amplifiers provide the following major advantages over conventional repeaters.
Figure 5 - Decay of Ions in a pumped-Er Fibre
significant equipment cost reductions
improved ability to upgrade
improved reliability brought about by the simplification of repeaters; and
higher bit-rates achievable
Using a power optical amplifier at the transmit end and a pre-amplifier at the receive end of a fibre system, it is now possible to achieve repeaterless transmission distances of 300km at data-rates of 2.4Gbit/s.
Solitons are rather avant garde in the sense that little has been written about them outside of the scientific community. However, the very future of fibre data transmission could lay in their hands. So, what are solitons? As we have seen above and in TW #3, even though data rates are often said to be infinite in monomode fibre, achievable rates are still limited to several Gbit/s and even then amplifiers are needed to achieve spans in excess of 100km. Solitons are the ultimate gee whiz technology, but promise to deliver data rates greater than 10Gbit/s with far fewer optical amplifiers.
Figure 7 - Pulse Widening Caused by Dispersion
It seems that John Scott Russell, a Scottish ship builder and engineer in the late 19th century provided the first description of what later became known as the soliton. He was riding his horse one day along the banks of the Union canal in Scotland when he observed that a 'solitary' wave created by a barge travelled for many miles without deterioration. He spent much of his remaining life trying to understand and explain this phenomenon. A soliton is a wave that does not broaden or lose its shape when travelling in a medium. In communications, a soliton is a light pulse of special shape propagating in monomode fibre.
In 1972 two Soviet physicists, Zacharov and Shabat, established the possibility of optical solitons when they showed that a particular non-linear differential equation (Schrödinger equation, if you can remember this from your university days!) produced a wave that did not dissipate and could recover its previous shape when subjected to external interference. An optical soliton is created in a fibre due to the interaction between two opposing factors. The first factor is that above 1300nm shorter wavelengths travel faster. As an optical pulse travels down a fibre, the longer wavelength components of the light pulse tend to fall behind. Thus the length of the pulse is increased by extending the trailing edge, so that the frequency at the leading edge is higher than at the trailing edge. This is called optical dispersion, as discussed in TW #3, and causes conventional optical pulses to broaden in proportion to the distance travelled, as shown in figure 7. The effect of this is to limit the data rates that can be achieved on monomode fibres.
The second factor is caused by amplitude non- linearity of the fibre. This results in increasing refractive index of the fibre with increasing pulse intensity. The practical result of this is that the phase or frequency of the pulse changes with intensity. The change is small but brings about major effects.
The soliton pulse is a bell shaped pulse as shown in figure 8 and has a particular shape. Each part of the pulse on the amplitude axis will undergo a frequency shift in the opposite sense to those caused by the dispersion in the horizontal axis. Therefore, solitons form when there is a balance between these two opposing factors, group-velocity dispersion and non-linearity.
Figure 8 - The Soliton Optical Pulse
Solitons have 'magic' properties, in that they will propagate without deterioration over many thousands of kilometres and as a single soliton is only 1 picosecond long (1 million millionth of a second) long they should permit realisable data rates in excess of 10Gbit/s.
Solitons are not a perpetual motion machine and are in reality subject to several limitations. However, it has been shown recently that solitons that have decayed can be revitalised, restored and relaunched following a stage of amplification. The real issue with solitons is not so much how far they will travel, as this is pretty much understood, but how closely can they be packed. The limitation to this seems to be the noise generated by the regeneration amplifiers. What occurs is that a generated noise photon has the effect of shifting the frequency of the entire soliton, effectively destroying the balanced nature of the pulse. Also, it seems that solitons can pass through each other and recover, but in a data communications role this would obviously lead to data corruption and is unacceptable. The best source of solitons is Er-doped fibre, the same as used in optical amplifiers.
The next generation of intercontinental systems such as TAT-12 across the Atlantic, will use optical amplifiers and make do with the limitations of conventional technology. By the end of the century, however, it is likely that long distance communication will dominated by solitons.
Figure 9 - A soliton based NOR switch (Source: Byte Magazine)
Solitons are far from being just an interesting laboratory phenomena, as the following examples will show.
Nippon Denshin Denwa (NTT) has used solitons to transmit data at speeds of 20Gbit/s (yes, twenty!) over a distance of 1000km, which they believe to be a world record. Their experiments, undertaken in 1991, demonstrated transmission at 10Gbit/s over a 1500km optical fibre loop. NTT used erbium-doped fibre amplifiers and believe that narrower solitons could be used to increase data rates beyond 1 terra bits (one million million million) per second. NTT expect to deploy this technology in 1994/5 time period.
AT&T and Bell
AT&T and Bell scientists have demonstrated error-free transmission of data using solitons at 5Gbit/s over 15,000 km.
There has been a lot of discussion in the last few years about the commercial possibility of optical logic. This, like the two previous subjects discussed, falls under the heading of photonics.
Figure 10 - Optical Signals in a Soliton Based NOR Gate (Source: Byte magazine)
As with optical regeneration, the need to convert optical data back to baseband electrical signals is tremendously cumbersome and slow. New techniques hold out the hope that it will soon be possible to execute logic at the optical level in the form of switches or binary logic such as NOR and NAND gates. Although this technology still lies in the realm of the laboratory, a most interesting application of solitons can be seen in the SOLITON NOR GATE. It doesn't take too much imagination to see what could be achieved if this approach to optical logic can be commercialised by the end of the decade.
So, how does this optical logic work? Again, it is based on erbium-doped fibre. Not only does this fibre enable the laser generation of high-power solitons, and amplify optical low-power optical data streams, but it can act as a logic gate as well! It is hard to believe that all this can be achieved from a piece of passive glass! Erbium fibre is an attractive medium in which to make all-optical logic gates, because such gates have an almost instantaneous response. By using long lengths of inexpensive fibre, low switching energies can be achieved. The technology is called soliton-dragging logic gates and can satisfy all the needs for logic gates or switches in an optical computer or digital communications data switch. It is possible to cascade gates, connect the output of one gate to many inputs of others and vice versa. It is also possible to perform the complete range of Boolean logic operations including addition and subtraction.
In soliton-dragging logic, the speed or propagation time of the control pulse depends on whether it propagates alone through a gate, or whether it is accompanied by another signal. Physically, when two pulses are coincident they 'chirp' and interact with each other by changing each other's centre frequency This causes a change in the velocity of the soliton because of the frequency dispersion explained earlier. Therefore, by looking for the output signal at a specific time it possible to differentiate between there being two inputs present or only one i.e. a logic NOR function.
The unusual schematic of a soliton-dragging NOR gate is shown in figure 10. The gate is made from two lengths of fibre, the first being typically 75 metres long and the second 350m (this is not exactly a miniature logic gate!). For the more technically minded, the control soliton is propagated in a vertical axis while the signal soliton is propagated in the horizontal axis. Hence, a polariser is required to separate the control signal from data signals. The gate is based on a synchronous clock, as seen with SDH, and is based on time-shift keying. That is, if the control soliton arrives within a specified time window the output corresponds to a binary '1'. If the control soliton in absent in the window the output is a binary '0'. In other words, the NOR function. This function is shown graphically in figure 9 above.
A soliton-dragging NOR gate is capable of operating at speeds of 200Gbit/s. This is around 1000 times faster than today's fastest silicon based logic!
Although quite physically bulky, this approach to super-fast optical switching is currently realisable and could be seen in commercial tele-communications switches before the end of the decade.
Photonics in General
Under the umbrella of photonics, come many other interesting subjects such as optical memories, optical backplanes in computers and telecommunications equipment, holographic applications, 3D storage on CD-ROM and many more. These will be looked at in more detail in future issues of Technology Watch.
Optical amplifiers, solitons, and optical switches are leaving the research and development environment and by the end of the decade data transmission and switching could be transformed. I say 'could' because like many other technologies commercialisation and cost reduction may pose insurmountable problems to exploitation. I would imagine though, that we all ought to start taking lessons in optics as silicon based transmission and switching eqipment could become a thing of the past all too fast!
I would like to thank STC Submarine Systems Ltd for some the information used in this issue of TW.
The information on Soliton switches was based on an article in the September 1992 issue of Byte. For further reading on the subject matter of photonics, this set of articles is highly recommended.