Telecommunications

Internet speed record shattered at 178 terabits per second

Internet speed record shattere...
A new internet speed record has been set at 178 terabits per second, using existing fiber optics
A new internet speed record has been set at 178 terabits per second, using existing fiber optics
View 2 Images
A new internet speed record has been set at 178 terabits per second, using existing fiber optics
1/2
A new internet speed record has been set at 178 terabits per second, using existing fiber optics
Lidia Galdino, lead researcher on the study
2/2
Lidia Galdino, lead researcher on the study

The fastest internet speed in the world has been clocked at an incredible 178 terabits per second (Tb/s) – fast enough to download the entire Netflix library in under a second. Engineers in the UK and Japan have developed new ways to modulate light before it’s beamed down optical fibers, allowing for much wider bandwidths than usual.

That new top speed is an insane feat. It’s 17,800 times faster than the current fastest internet connections available to consumers – 10 Gb/s in parts of places like Japan, the US and New Zealand. Even NASA can’t compete, with its 400 Gb/s ESnet.

It also leaves other experimental devices in the dust, including a photonic chip developed in Australia that clocked a still-impressive 44 Tb/s just a few months ago, and beats the previous record holder – a Japanese team with 150 Tb/s – by almost 20 percent.

“While current state-of-the-art cloud data-centre interconnections are capable of transporting up to 35 terabits a second, we are working with new technologies that utilize more efficiently the existing infrastructure, making better use of optical fiber bandwidth and enabling a world record transmission rate of 178 terabits a second,” says Lidia Galdino, lead researcher on the study.

Lidia Galdino, lead researcher on the study
Lidia Galdino, lead researcher on the study

To hit these speeds, engineers at University College London (UCL), Xtera and KDDI Research developed new technologies to essentially squeeze more information through the existing fiber optic infrastructure. Most are currently capable of a bandwidth of up to 4.5 THz, with some new technologies approaching 9 THz. The team’s new system, however, raises the bar to 16.8 THz.

To get this much extra “room,” the researchers develop new Geometric Shaping (GS) constellations. Basically, these are patterns of signal combinations that alter the phase, brightness and polarization of the wavelengths, in order to fit more information into light without the wavelengths interfering with each other. This was done by combining different existing amplifier technologies into a hybrid system.

Perhaps the best news is that because it uses the fiber optic cables already in place in many parts of the world, this technology could be integrated into existing infrastructure relatively easily. Instead of replacing miles and miles of cable, it would only require upgrades to the amplifiers, which appear every 40 to 100 km (25 to 62 mi) or so.

The research was published in the journal IEEE Photonics Technology Letters.

Source: UCL

12 comments
TorATB
I still got my 3.5Mb/s copper connection... Sucks
sidmehta
Now we need 3D hologram movies so these speeds are too slow :) and much faster speeds have to be invented. To handle such vast amounts of data we need quantum computers, which are on the way. We'll have the best whiz-bang technologies while we fight over toilet paper and wage wars!
Brian M
"fast enough to download the entire Netflix library in under a second"
and still nothing worth watching.....

Jarid Orgeron
False - You would need about an 800 Tb/s connection to download the approximate size of an end user Netflix library of around 100TB in one second.
PAV
I was not aware that you had to have the amplifiers every so many miles. How does that work?
Daishi
@PAV some modulations are capable of longer reaches with a weak signal but other higher bandwidth modulations need a strong clear signal to decode. Something like NRZ where the light is just on or off you can understand at long distance with a weak signal but you can't modulate very much data with just NRZ. Something like QAM16 is a more common modulation today that is a pretty good sweet spot for the distances you can reach with it. You can look up how EDFA amplifiers work for a better description but essentially photons are pumped into the fiber line at a different wavelength than the signal to boost the power of the signal-carrying photons. Part of the issue with amplification is that it also tends to amplify the noise as well so you can only get 7-10 amps on a line before you can no longer tell the difference between signal and noise. The problem with really high modulations is that they are often only useful across really short distances. One simplistic example is if I yell at you it means something different than a whisper and you can tell them apart easily, but if I speak to you at 4 different volumes and I want them all to mean something different maybe you can tell them apart, but if I communicate to you at 100 different volumes you need a really clear signal and precise instrumentation to tell the difference between volume 98 and volume 99 and that becomes harder and harder the farther away you are and more faded my voice becomes. That's amplitude modulation, frequency modulation would be doing the same thing with the pitch of my voice. You combine different kinds of modulation and you can send a lot of data. Any variable you can change that can be understood at the other end can be used to communicate more information, amplitude, frequency, phase, etc. Results like this come with a large asterisk next to them and the most important of which is that only some frequencies (mostly C-band) are suitable for sending data over fiber at distance. Their test used S, C, and L bands together to achieve the result.
Ray6969
Nice for sure but until it's available to the general public it's just a neat invention without much interest to the common person. Science types like myself will find it interesting I'm sure, I know I did.
Catweazle
10 Mb/sec fast? Even out here in the Yorkshire Dales I am getting 70 Mb/sec according to the British Telecom line tester.
ljaques
Boy, that sounds good. Centurylink still has us under a 3.5Mb/s ADSL noose in Oregon.
Knut
@Daishi @PAV - There is no "delay" in light, and you cannot modulate the frequency. Please understand that US universities are decades behind. Light can be modulated in wavelength only - change the density of the glass, and bounce off a mirror.
They use the wavelength to multiplex and send multiple "channels" on one fibre. Then at some point, you send more than a programmable chip can intercept - it can just execute at 3GHz clock cycles - but a prism or mirror runs faster. So many runs at the same time and individual "channels" cannot be intercepted. The fibre is put in the ground and tuned to compensate for variation in density when it was produced.
The same fibre runs the same distance, not quite from USA to China, but it is hundreds of km/miles. The light is one beam, and we see the wavelength as the different colours. If a hundred people can share the same fibre at the same speed, the cost of pulling the fibre is shared by everyone.
Europe knew that the speed of light was the same for everyone - socialistic and shared. So they made standards for multiplexing on the fibre and has for decades helped China and everyone to develop this technology. I had some consultants working on "inference" between the channels some years ago, they wanted this in-between channels to be used for "my turn/ your turn" kind of information.
These fibres run the internet outside the USA and is used worldwide for mobile phones. These are the international standards that the FCC will not approve of because they want to secure a place for a US company to "discover" it and "patent" the "technology" so that everybody can pay to the FCC for isolating America.
What none of you detected is that at 3000GHz, the light travels less than an inch, about 1/10 of a millimetre and that is very difficult to make a circuit with programmable circuits this small. The ITU assumes that 4KBytes are collected and routed per slot that then can be routed as those that have configured the fibre wants the 4K "packets" delivered. This is an "increment" or "decrement" on your mobile phone bill. What is described here is that STM1000 is possible, and even above this. This means that the operators that have pulled the fibre can by adding ITU equipment, put 1000 more on the same fibre and get paid many times over for sharing the bandwith - every one of the hundreds gets their own private fibre capacity, uplink and downlink can be 2 separate channels.
This is politically motivated, and not limited by physics. The light runs socialistically just as fast for everyone and can be shared.