Optical amplifiers - a telecommunications revolution
At EPSRC ITeC '96: Engineering and Physical Sciences Research Council ITeC Conference, United Kingdom.
01 Oct 1996.
Full text not available from this repository.
The march of optics into telecommunications transmission: There is nothing new about optical communications. The 'beacon hills' all over the UK once communicated from London to Paris by visual signals. The use of electricity to develop telephone networks moved attention to that technology for a time. But optics is now capable of delivering far better performance at lower costs than coaxial cables. For instance, one fibre strand is capable of carrying at least 1500 digital TV channels while a bandwidth about as high as 3 TeraHertz (THz) might be available.
Traditionally, an optical communications system worked by sending a light from a source via an electrical modulator down a length of fibre. At the receiving end, the signal goes through a demultiplexor and out again to the next section of fibre via a repeater amplifier. Electronic amplifiers were needed about every 100km or so because of signal attenuation, and since these are slow, the full potential of the 3THz can't be exploited.
All this changed in the mid-1980s with the appearance of optical-fibre amplifiers, such as the Erbium-Doped Fibre Amplifier (EDFA). These devices take input signals which may come from 10,000km away and perform enormous amplification, but without the reshaping and retiming of pulses as happened with electrical amplifiers. With the resultant 'optical transparency', any form of signal can go through a global network without cross-talk. This allows, for example, for analogue working in some channels and digital operations in others.
Signal dispersion remains a major problem on the global network. Dispersion first emerged as a major transmission issue with the laying of the first transatlantic cable in 1866, when dispersion caused such bad signal smearing that it was first thought the cable was broken. The problem was eventually solved by using coil equalisers along the cable's length.
Modern dispersion problems are exacerbated because optical amplifiers work at a wavelength of 1.55 microns, whereas the installed network was designed to operate at 1.3 microns. This limits the transmission span to about 60km at 10Gbt/s. To overcome the problem, we have a choice of either developing a viable 1.3 microns amplifier, if that is possible, or developing dispersion-compensation techniques for the EDFA approach.
There are a number of ways in which such compensation can be achieved, such as the optical equalizer device a couple of centimetres long developed at our Southampton Centre. Once such compensation techniques become widely available, optical systems will be able to deliver the telecommunications dream of something like 3THz operation with signal transparency and no dispersion. No other information transmission medium can match this.
The impact and potential of optical systems:
A few years ago, a BT experiment showed that optical amplifiers could enable a single video server to distribute sixteen 2.488Gbps per channel to nearly 44 million subscribers over a distance of 527km. This could reach just about everyone in the UK. And an experiment by NTT of 160Gbps transmission over 62km suggested that it would be feasible one day to push the equivalent of the total maximum communications traffic in the USA down one fibre.
To gauge what all this means, consider the complete American Library of Congress, which consists of about 25 terabits of information. It would take around 4, 000 years to send this by fax to Japan; ten hours by 10Gbps optical fibre; and only a minute if the full capacity of optical fibre technology is exploited. We will need to give a lot of careful thought to what we do with this growing abundance of transmission space created by optical systems.
In addition, optical fibre technology is revolutionising the telecommunications industry. For example, optical networks are seriously questioning the need for satellite communications. They have also cut the cost of transmissions to the point that it now costs a telecoms company about the same to handle a call from Leeds to New York as from Leeds to Manchester, despite the different rates charged to customers. It is also interesting to observe the enormous integration in the market of entertainment media and telecommunications activities, such as the way cable TV companies are now also becoming telecom operators.
Ultra-fast optical switching in processors consumes a great deal of power and introduces a large pipeline delay, so in my view we are not going to see general-purpose optical computing. But we are already seeing telecommunications multiplexors with optical-signal routing, because a pipeline delay doesn't matter when a signal is coming from thousands of miles away.
Some people have also been looking at single- mode fibres which can produce ultra short picosecond pulses far better than you can do with electronics. However, a fibre of over 6km is now needed as a picosecond of optical demultiplexor, incurring a pipeline delay of 5 microseconds per kilometre. Optics is far behind silicon when it comes to integration, although the first steps in this direction are underway. One such device which will be a vital aid for distributing signals from 'fibre to the kerb' to many homes in an area is an 'optical splitter' that divides a signal by a hundred and amplifies it at the same time.
Eventually, optics could follow the example of silicon in going from discrete components to more integration. In the next 25 years we are likely to see growing use of optical amplifiers, faster pulse generation and transmission and more all-optical switching and routing. Such routing capabilities could allow a subscriber to choose the destination of a signal, rather than relying on a switch operation by a telecommunications company. There will also be more and more active optical circuits and sensors on a chip and more use of optics inside computers and in inter-chip operations.
Actions (login required)