History of the Atlantic Cable & Undersea Communications
|Happy Birthday EDFA
Happy Birthday EDFA
The 4th May 2011 was the 25th anniversary of the opening of the first international fibre optic submarine cable system (UK-Belgium N0. 5). However, perhaps of much greater significance and importance to our industry is the fact that in May this year, a single component that has probably had the greatest impact on optical telecommunications transmission and the growth of the Internet, joins the ‘Quarter Century Club’. The Erbium Doped Fibre Amplifier (EDFA) was first demonstrated by Professor David N Payne and his team at Southampton University (UK) in 1986.
For those readers who are not technically minded: in simple terms, the EDFA consists of a length of optical fibre doped with ions of the rare earth element erbium. Within the fibre the optical transmission signal wavelengths are mixed with a high powered signal from a pump laser. This excites the ions in the erbium and they give up this energy in the form of additional photons in phase with the incoming data signal, thus amplifying it.
A Simple Erbium Doped Fibre Amplifier
So why is the EDFA so important?
Prior to the invention of the EDFA, optical transmission systems had operated over long distances by the use of electro-optic regeneration. As an optical pulse travels along a fibre it is attenuated and loses it shape; eventually the information contained in the pulse will be lost, unless before this occurs the signal can to be boosted. Initially, the only viable method of doing this was to detect the incoming signal with a photo diode, convert it to an electronic signal, then use that signal to switch a laser, thus sending out a new optical pulse along the next length of fibre. This type of system had the limitation of only being able to work with one signal (wavelength) and the system had to be designed to operate at a specific bit rate.
In contrast, the EDFA offered direct amplification independent of the signal bit rate, and had the ability to allow greater spacing between repeaters. The EDFA is also a much simpler device than the regenerative circuitry of a repeater and so it is far more reliable, which is of particular importance for submerged repeaters.
From its initial invention, it took several years for Payne’s group in Southampton and a parallel development team at Bell Labs, under Emmanuel Desurvire, to produce a reliable EDFA that could be manufactured in volume. During their experiments they found that the EDFA could simultaneously amplify signals at two or more different wavelengths (Wave Division Multiplexing or WDM), which was not possible with regenerative systems. So EDFAs appeared to offer the opportunity for the design of much higher capacity systems than could be achieved with regenerators.
The first transatlantic fibre optic system was TAT-8 (1988). This was a regenerative system operating at 1,310nm and a line rate of 240Mbits/s. In 1991 TAT-9 went into operation; this was still a regenerative system but operated at 1,550nm and a line rate of 560Mbit/s. The first transatlantic optically amplified systems were TAT-12 and TAT-13 in a ring network; these systems utilised a single wavelength of 1,550nm with a line rate of 5Gbits/s and opened for service in 1996.
The invention of the EDFA coincided with the start of the dot com boom and the deregulation/liberalisation of the telecommunications industry in the USA and the UK. Competition between telecoms companies, the increasing demand for capacity from the Internet, and the willingness of banks to fund submarine cable projects, created an environment that sent the industry into a boom period. With so many traditional carriers and start-up companies wishing to build new cable systems, suppliers looked to find ways of differentiating themselves. The EDFA gave them the opportunity to develop and offer systems with more and more capacity on a single fibre pair. Because of the EDFA the concept of the “transparent pipe” became popular; the idea that the capacity of a fibre system is only limited by the equipment connected to each end. This is, of course, an over-simplification; the number of wavelengths that can be transmitted and amplified over long distances is dependent on the bandwidth of the EDFA and the ability to separate wavelengths. The bandwidth is a function of the properties of the erbium doping ions, the structure of the fibre, and the power and wavelength of the pump laser. All submarine systems were (and still are) designed to have an ultimate or ‘Design Capacity’ based on the technology available at the time, and generally they are initially equipped at a much lower capacity level, allowing for growth over their theoretical 25-year lives.
WDM was quickly developed to offer 16 wavelengths per fibre pair, and with the ability to space wavelengths even closer together, Dense Wave Division Multiplexing (DWDM) was developed for terrestrial systems, and then quickly taken up by the submarine cable industry.
Dense Wave Division Multiplexing
In a relatively short timescale, when one takes into account the rigorous reliability requirements of submarine system, by the year 2000 the available capacity on a fibre pair for an optically amplified system had moved from one wavelength (λ) at 5Gbit/s to an industry standard offering of 64λ x 10Gbit/s = 640Gbit/s. Of course the total capacity of a submarine cable is a function of line rate, the number of of wave lengths, and the number of fibre pairs in the cable. For repeatered systems, the number of fibre pairs is constrained by the number of amplifiers that can be accommodated in the repeater housings and that can powered through the cable. From its inception, the repeatered fibre optic system model had been built around a maximum of four fibre pairs per system, but during the boom, design and development was undertaken for six and eight fibre pair repeatered systems. For repeaterless systems there was no such constraint and one system was installed between the UK and Belgium which contained 96 fibre pairs. The EDFA also played a major role in the development of repeaterless systems, extending the distance that could be spanned through the use of transmit and receive amplifiers plus Remote Pumped Optical Amplifiers (ROPA). This made fibre connectivity possible to islands and remote locations where the cost of repeatered systems could not be commercially justified.
In the year 2000 the submarine cable industry was on the crest of a wave, buoyed by what in retrospect was an insane optimism that the exponential growth in capacity demand then being enjoyed would go on forever. For many reasons, the boom had created a market where in reality it was the provision of new capacity rather the demand for capacity that was spiralling. It had been expected that bandwidth-hungry applications such as video streaming, 3G mobile services, Internet data, etc. would create the demand for huge capacities to be available on the major backbone and Internet routes, and although demand did continue to rise at a steady rate, history has shown that the forecasts for growth were excessively optimistic. The bubble burst, and for the next five years very few new systems were required, as the continuing growth of demand was taken up by the commissioning of previously unlit capacity on existing systems. The transparency of the EDFAs also allowed, through advances in technology, systems to be upgraded far beyond their original design capacity.
As the existing systems filled up and new technology allowed them to carry more than their original design capacity, the international networks became vulnerable to single point failures because of the lack of route diversity. The need to make existing networks more robust started the recovery of the industry.
Today, submarine cable systems can support > 100λ at 10Gbit/s and multi-wavelength 40Gbit/s transmission on a single fibre pair. Repeaterless systems of up to 400km in length can be accommodated with ease, and more ‘heroic’ systems of 400 – 500km in length are possible. This is all because of the EDFA.
It is a sobering thought that when you downloaded this article, the data stream probably passed through several hundred of these devices on its journey from our server to your computer.
Article text copyright © 2016 Stewart Ash
Last revised: 22 October, 2016