News Japanese Institute breaks optical fiber speed record with 22.9 petabits per second — 1,000 times faster than existing cables

[Admin]

NCIT collaborated with two other institutions to mark a new world record capable of handling 1000 times more traffic than existing worldwide internet networks

Japanese Institute breaks optical fiber speed record with 22.9 petabits per second — 1,000 times faster than existing cables : Read more

 

[George³]

I read, read, read and didn't figure out.
It's as if this article is assembled from a title and separate sentences written by different people and contradicting each other throughout the text. They didn't get along. One claims all the speed through a single optical fiber, and another that it is through an entire cable bundle of several dozen optical fibers.

 

[PEnns]

Japan? The universities in the article are in the Netherlands and Italy.

Anyway, in many parts of the US peoples' only access to the internet is a DSL cable....they can only dream of such speeds! And we're not talking about rural areas only, but places just a couple of miles outside major cities!

Yeah, I lived till 6 months ago in one where 1 mbps was the top speed on a good day! Download speed was about 700 kbps.....and that amazing speed cost "only" $90 / month.

 

[HaninTH]

Anyway, in many parts of the US peoples' only access to the internet is a DSL cable....they can only dream of such speeds! And we're not talking about rural areas only, but places just a couple of miles outside major cities!

Yeah, I lived till 6 months ago in one where 1 mbps was the top speed on a good day! Download speed was about 700 kbps.....and that amazing speed cost "only" $90 / month.

It's a tragedy of our greedy capitalist overlords that they hold back our internet access so drastically. There's unlit fiber all over the place, but these maniacs refuse to utilize it because they can't be ensured a captured revenue stream until they hit a certain density threshold.

Maybe we need to revisit the notion that corporations should only exist to generate revenue and add a little customer support mantra into the mix so that there's a reasonable blend of revenue to service.

But this is a pipe dream....

 

[JarredWaltonGPU]

Japan? The universities in the article are in the Netherlands and Italy.

It's literally the first sentence of the abstract, as well as the article:

"Researchers from the National Institute of Information and Communications Technology (NICT, President: TOKUDA Hideyuki, Ph.D.), in collaboration with the Eindhoven University of Technology and University of L’Aquila demonstrated a record-breaking data-rate of 22.9 petabits per second using only a single optical fiber, which was more than double our previous world record of 10.66 petabits per second."

So yes, it's a Japanese Institute, in collaboration with Netherlands and Italy universities.

 

[COLGeek]

Are these the new arteries of the AI overlords that await the destruction of humanity that so many are losing their minds over?

 

[George³]

Obviously the text will need to be revised, otherwise some of the statements do not match the graph included and that a single optical fiber has a performance of 0.3 to 0.7 petabits.

 

[TJ Hooker]

Obviously the text will need to be revised, otherwise some of the statements do not match the graph included and that a single optical fiber has a performance of 0.3 to 0.7 petabits.

Which exact statements? It's a single cable with 38 cores. Each core can transmit between 0.3 to 0.7 Pb/s, with the average per-core data rate being 0.6 Pb/s (22.9 Pb/s divided by 38).

 

[George³]

Which exact statements? It's a single cable with 38 cores. Each core can transmit between 0.3 to 0.7 Pb/s, with the average per-core data rate being 0.6 Pb/s (22.9 Pb/s divided by 38).

Hahaha. The problem may also come from a misunderstanding of terminology. The difference between a single optical fiber and the cable used is apparently incomprehensible. The claim from the title and from some sentences in the text is as if someone split the cable lengthwise, removed just one of the fibers and let all the traffic go through it.

 

[TJ Hooker]

To put it in plain words, this new infrastructure could handle up to three times the amount of data traffic through these cables once deployed, which is 1,000 times more than the currently deployed optic cables.

I'm not sure where the "three times" came from. Maybe a typo/misunderstanding of the following statement from the source article?

"[...] ultra-large-capacity optical fibers will be needed in the future, where the data traffic demand is expected to increase by 3 orders of magnitude (x1,000 times)"

 

[shpankey]

Limitations like this have nothing to do with FoC itself but the equipment transmitting and receiving on either end.

 

[JarredWaltonGPU]

I'm not sure where the "three times" came from. Maybe a typo/misunderstanding of the following statement from the source article?

"[...] ultra-large-capacity optical fibers will be needed in the future, where the data traffic demand is expected to increase by 3 orders of magnitude (x1,000 times)"

I believe it's 3X the throughput of other designs using the same cable. Those cables (38 core) are not yet deployed, so it's 1,000 times the throughput of currently deployed systems. From the paper:

"The measured transmission capacity for each core ranged from ~0.3 to 0.7 petabits per second leading to a total transmission capacity of 22.9 petabits per second. The achieved data-rate includes an overhead for an implemented forward-error correction code with the demonstration showing up to 24.7 Pb/s can be achieved with better optimized coding. This is more than 1,000 times the data-rate of currently deployed optical fiber communication systems."

That's right before the next paragraph that you quoted in part:

"While uncoupled four-core MCF is suitable for early adaptation, further improvement of the telecommunication infrastructure using ultra-large-capacity optical fibers will be needed in the future, where the data traffic demand is expected to increase by 3 orders of magnitude (x1,000 times). This study demonstrates the first successful combination of multi-band WDM and SDM employing a multicore multimode fiber, which is key to the realization of future ultra-large-capacity optical fiber communication networks."

 

[bit_user]

Anyway, in many parts of the US peoples' only access to the internet is a DSL cable....they can only dream of such speeds! And we're not talking about rural areas only, but places just a couple of miles outside major cities!

Seriously?? This is not for home internet! It's for long-haul backbone links, like transoceanic cables and links between major datacenters.

 

[bit_user]

It's a single cable with 38 cores. Each core can transmit between 0.3 to 0.7 Pb/s, with the average per-core data rate being 0.6 Pb/s (22.9 Pb/s divided by 38).

It's also probably worth pointing out how they reach 114 spatial channels.

​

38 fibers x 3 modes = 114 spatial channels.

 

[TomSteve22]

The researchers achieved this breakthrough by combining the latest research technologies with Space Division Multiplexing (SDM) and Wavelength Division Multiplexing (WDM). SDM uses multiple multicore fibers and various transmission technologies to work with over 100 spatial channels, which are then combined with multi-bandwidth WDM. The researchers merged this 38-core, three-mode cable through a multi-band compatible MIMO receiver.

This article feels a bit like it was written with AI, especially this portion. Research and its permutations are used three times in a single paragraph.

 

[bit_user]

This article feels a bit like it was written with AI, especially this portion. Research and its permutations are used three times in a single paragraph.

Let's please focus on the content. The articles are churned out at such a rate and for an audience small enough that I think we have to accept they won't read like your leading print newspaper of choice.

What I care about is whether the articles get the details right and are worth reading, overall. Sadly, even this seems a high bar and one that's routinely missed on at least one other popular tech site I have sometimes skimmed in the past. I say "skimmed", because the writing on that site is/was so bad it often hurt my head to read the articles in full.