A speed revolution in data transmission - Real or Hype?
Alright. Maybe some of you noticed, maybe some of you didn't. Early in April this year there was an article, (or rather several) about how Bell Labs in collaboration with Aston University managed to send data at a rate of 301 Tbps (Terabits per second). One of the articles (the source for some of the others, I believe) can be found here:
As you can see - Headline said it's 4.5 MILLION times faster than the average broadband. I was intrigued, because if this is true it would truly revolutionize the internet. Such an increase in throughput would mean that it's the same factor of difference as between a 2400 baud modem from the end of the 1980s to a 10Gb connection! It's simply a ludicrous speed increase and it would totally flip the script on what is the bottleneck in our internet communication. Suddenly the home networks would be the slow parts, and our computers capacity to receive data over old-fashioned Wi-Fi or Ethernet would be what limits our consumption. Sounds to good to be true, right? I thought so, so I decided to actually read the article thoroughly instead of just going slightly insane thinking about all the possibilities this could bring.
What I found on closer inspection is that what is compared in the article is the speed of the average home connection in UK (69.4Mbps) with that theoretical top speed of 301Tb. Such speeds are more likely to be in the backbone of the really large ISP:s that already operate with extremely high capacity, rather than individual homes. So - it's a bit like saying "If we could connect your faucet directly to the city main water line you could fill your pool in just over a minute". While theoretically true, it doesn't mean that you could use all that water to yourself, or that it would be in the interest of the city to allow you to. You simply can't push 565 liters per second through an ordinary faucet in your home - and you can't push 301Tbit through DSL or 5G, or even if you have fiber. ISP:s often use Passive Optical Networks (PON) where you in fact share a single fiber on 32 to 64 households, sometimes even more. ISP's often use oversubscription where the combined bandwidth sold to the households exceed the capacity, simply because households rarely utilize their capacity to the max simultaneously.
So - is this all just nonsense hype then?
I'd say no - here are the technical parts that are actually kind of cool about the speed increase:
If you check the data given in the NICT link below, you'll see that the theoretical speed is even higher today. Through adding the O and U band, they increased the capacity to 402 Tbit over 50 km of fiber.
BUT - you'll also see that the capacity for using C, L and S bands is 226 Tbit in the same test. Most ISPs to my knowledge today use C and L band because they are suitable for long range transmission, with a theoretical figure in the same kind of test at about 130 Tbit if my amateur calculations are correct.
So that's the speed the really large players backbones operate at today, like Google. In fact, their Dunant Cable between France and USA is capable of 250Tbps, which means that 402Tbps is less than twice that. The Dunant Cable is fairly new though (2021) and using multiple fiber pairs to transfer data.
As you can see the touted 4.5 million times speed increase is in fact not even close. It is -at best- around a 4x increase over current speeds. Sadly this means that you will NOT be able to download 335 8k movies per second, or that we can ignore the constraints of our internet connections like packet-loss etc.
Still though, it would be a a significant increase in the backbones of the internet if all the bands were to be used, especially considering it's done with existing fiber-optic infrastructure. Quadruple the throughput of data in the same infrastructure is not to be sneezed at.
But - also consider that these backbones carry ALL the data, it's not just our Netflix movies and doom-scrolling on YouTube it's meant to handle. It is also IoT with Digital Twins, our new AI services, remote operations of machines, edge computing (distributed data processing) and much more! We are constantly increasing our bandwidth consumption as a society, so odds are this heightened capacity will not be all that noticeable for the end household consumer, if at all.
领英推荐
So - in summary, it's good for the ISP:s who can utilize their existing infrastructure more effectively and possibly increasing ROI on it, but it's unlikely to be a speed-revolution for us as end-consumers. The speed increase is certainly impressive considering the constraint of using existing fiber-infrastructure, but if we remove that constraint however, it turns out there are more promising technologies on the horizon.
Wait what? Is there FASTER stuff?
Yes - Using other technology than existing fiber, the record transmission speeds recorded is at 22.9 Pbs (Petabits per second). That used multicore fiber, and the speed is around 57 times faster than using ALL the bands listed in NICTs tests in existing fiber, and around 200 times faster than the big ISPs capacity today. Though perhaps not ready for commercial use on a large scale, I think this is probably the likely route that we must take in the long run, to be able to supply the bandwidth necessary for our needs as a society in the future. Here's a chart illustrating the difference in speeds over time. Keep in mind that this is a logarithmical scale, if it was linear the 22.9Pbit would simply be through the roof while everything on the left would be in essence invisible in comparison.
So - why don't we just simply replace the old with new stuff?
Even if we do that, (aside from the astronomical costs involved and that it's probably not ready for large scale deployment) - it's the 'last mile' that limits our consumption as consumers. We simply don't have the infrastructure to utilize those high-speed transmission technologies all the way to our homes yet, and it's perhaps more reasonable to share the bandwidth than to have your own dedicated connection that is mostly idle?
Initiatives like Fiber to the Home (FTTH) and advancements in throughput in 5G will surely make it possible for higher speeds, if you are willing to pay for it. Perhaps we can hope for slightly higher speeds at lower prices given a functional market economy, but I don't think we should expect a revolution.
Some interesting links to read if you want to go down that rabbit-hole:
What do you think? Will we experience no-latency "all-you-can-eat" internet in our lifetimes? Will it be generally accessible or limited to just Urban areas? Will supplied speeds to household from the ISP's ever exceed the capabilities of our home-networks, leaving our homes as the bottlenecks?