Real-time transatlantic submarine cable capacity and wavelength records broken - again

• Infinera field trial hits 30 Tb/s capacity and first 700 Gb/s wavelength in deployment test
• ICE6 Optical Engine used on the MAREA cable with remarkable results
• Hyperscale ICPs now consume 70 per cent of all submarine cable network capacity and are greedy for more
• Route to petabit capacity within 5 to 10 years now open

Talk about "If you build it, they will come"! Once upon a time (actually less than a generation ago) there was a lot of market sentiment and analyst comment that far too much cable was being laid and that the majority of it would lie there in the depths, dark and redundant, because there would never, could never be enough traffic to fill the perceived over-capacity. How wrong that was. Today upwards of 99 per cent of all the data traffic crossing the world's oceans is carried by subsea optical cables and over the past 30 years they have become utterly essential to the global economy.

Submarine fibre-optics cables are very robust and highly reliable, the technology is tried and tested and their carrying capacity and speed of transmission continue to increase. One of the favourite maxims of Lieutenant Commander Montgomery "Scotty" Scott of the Star Trek science fiction TV series and films was,"Ye cannae change the laws of physics. Captain", but Infinera of Sunnyvale in Silicon Valley, well-known maker of optical transmission equipment with more than 2000 patents in its portfolio including optical transport and the virtualisation of optical bandwidth, is one of few companies pushing the envelope to its extremes as the race to provide petabit data rates within the next five to 10 years accelerates.

Most recently Infinera has been conducting a field trial over the MAREA transatlantic cable utilising its ICE6 Optical Engine and has achieved two remarkable world record results by breaking both the 30 Tb/s transatlantic capacity and the first 700 Gb/s wavelength. To understand quite what that means and why it is important it is necessary to be aware of the difference between the "hero result" of a test and the "deployable result" of a test.

Basically, the 4,000 mile-long MAREA cable connects Spain and the US. It is the highest-capacity submarine cable in the world, designed to carry 200 Tb/s. It is funded and owned by Facebook and Microsoft but was built and is operated by Telxius, which is a subsidiary of the Spanish national telco Telefónica. It is made up of eight pairs of fibre-optic cables bundles and went live in February, 2018.

MAREA, from the outset, was designed for very high-performance coherent transmission (in essence, coherent optical transmission utilises uses modulation of the amplitude and phase of the light, as well as transmission across two polarisations.This enables maximal transport of more data through the cable than would be possible without it), and uses large-area, low-loss optical fibre based on a pure silica core with a short 56 kilometre spacing between amplifiers which gives an excellent optical signal-to-noise ratio.

When it entered service its capacity and capabilities exceeded all of the other contemporary operational transatlantic cables combined. MAREA is now regarded as the benchmark against which all other transatlantic cable systems and infrastructure vendors are measured and compared.

So, a "hero result" is that obtained by not taking account of any margin for actual deployment of real-world traffic under real-world conditions, whereas the "deployment result" is that achieved where a margin is allocated for real services to be deployed at a particular data rate or total capacity. Infinera's MAREA hero results over the 6,640 kilometres between Bilbao in Spain and Virginia Beach, Virginia in the US were 30 Tb/s of total capacity over a single fibre pair and 700 Gb/s data rate per wavelength. The "deployment result" over exactly the same distance was 28 Tb/s of total capacity over a single fibre pair and up to 650 Gb/s data rate per wavelength.

In situ, submarine network technologies are ideal for experimentation to push boundaries of optical performance and for years now Infinera has been an expert at tweaking subsea fibre capacity to get more out of it than might seem possible given design constraints. Basically the company has, so far, always been able to find more gas in the tank even when the fuel gauge is hovering on empty. That's most important given that huge, hyperscale ICPs (Internet Protocol Communications Platforms) such as Facebook, Google et al now gobble up getting on for 70 per cent of all international subsea network capacity - and growing. 

Coming up against the Shannon Limit

The Shannon Limit is not about the number of salmon a fly fisherman can legally pull out of Ireland's longest river, but refers to the fact that the 28 Tb/s over a single fibre pair including commercial margins achieved in Infinera's ICE6 tests (up from the 24 Tb/s achieved in earlier tests with ICE4) is butting up against the Shannon Test, named after Claude Edward Shannon who, back in 1949, devised a theorem defining the maximum data rate over a communications medium (such as wireless, coax, twisted pair, fibre etc.) in the presence of noise without incurring transmission errors. Basically it means that the higher the signal-to-noise (SNR) ratio and more the channel bandwidth, the higher the possible data rate.

Every fibre pair has its own Shannon Limit and in the ICE6 tests the parameters were set to allow Infinera compensation algorithms to push performance right up against the Shannon limit. It was found that operating at a lower baud rate provides for better spectral efficiency with the result that an optimal 450 Gb/s per wave was needed in order to achieve 28 Tb/s in total.

What this means in practice is that as operators of subsea networks submarine network operators start to fill their cables, they could operate at high data rates that, while less spectrally efficient, are more cost effective on an interface basis. Then, as the cable utilisation increases, they can switch to a lower data rate and higher spectral efficiency and add transponders to provide the capacity they need.

With ICE6, the higher the data rate per wave, the better the cost per bit while fewer transponders are needed and less electrical power is consumed. The 28 Tb/s option with ICE6 reduces the network element count by 6o per cent as against the boxes previously required for 24 Tb/s, a huge improvement.

Looking to the future the trend in new submarine cables is towards space-division (or spatial-division) multiplexing (SDM). In the past, the goal was to maximise the amount of bandwidth per fibre pair, more wavelengths, and a higher bit rate per wavelength. Now the emphasis is on more fibre pairs in a single cable.

Traditional cable has four to eight to eight fibre pairs and that is pretty much the maximum in traditional subsea cable, However, with SDM, cables will have 12 to 16 fibre pairs (and, in due course, even more). The calculation is simple, more fibre pairs equals more cable capacity, for example, traditional submarine cables with six fibre pairs at 20 Tb/s each gives a total capacity of 120 Tb/s while SDM subsea cable with 16 fibre pairs at 16 Tb/s each provides total capacity of 256 Tb/s. SDM is regarded as a viable roadmap to 1 petabit trans-atlantic cable by or before 2030. Carriers will be pleased. So will Infinera.