If there’s one protocol that networkers are saturated with on a daily, if not minute-by-minute basis, it’s TCP/IP.
Well guess what? The now-aging TCP/IP might not be around for much longer.
Researchers at Aalborg University in Denmark, in association with MIT and Caltech, reckon that the Internet can be made faster, and more secure, by abandoning the whole concept of packets and error correction. Error correction slows down traffic because the chunks of data, in many cases, have to be sent more than once.
The researchers are using a mathematical equation instead. The formula figures out which parts of the data didn't make the hop. They say it works in lieu of the packet-resend.
Janus Heide, CEO of Steinwurf, a commercial enterprise set up to promote network coding, of which this is a part, and Frank Fitzek, a Professor at Aalborg University’s Department of Electronic Systems, draw an analogy with cars on a road - their technique is like pumping cars into an intersection from all directions, without the cars having to stop for each other. Traffic flows faster because there are no red lights.
The key to the process, they say, is a network coding and decoding element called RLNC, or Random Linear Network Coding. RLNC is the technology that they've patented and wrapped into C++ software at Steinwurf called Kodo. Steinwurf plans to sell its technology to hardware makers.
Results have been outstanding, according to the group. They used a four-minute video and reckon that it was downloaded five times faster than it would have been with other technology. The video also didn't stutter, unlike the original clip, which was interrupted 13 times. Buffering, as any Netflix-adopting cord-cutter knows, is the scourge of the Internet user today.
The tests show speeds five-to-ten times faster than usual. Fitzek says the technology can be used in satellite, mobile, and normal Internet communications.
All sounds good, right? Faster Internet, no more TCP/IP. Well, what’s a bit complicated is following along with the mathematics and coding. As you might imagine, it’s a bit more complicated than the common-or-garden TCP/IP stack in your average Windows Networking dialog box.
Conveniently for us, Fitzek has posted a YouTube video of a lecture from the 2014 Johannesberg Summit — that’s a future-of-wireless shindig — in which he attempts to explain it all.
There are some clever snippets related to the history of packet and networks, in relation to point-to-point telegraph, and mesh telephone systems. Plus, there’s a lively nugget on how he believes communications is really storage, because a buffered packet is, in fact, stored.
It’s all good stuff, and worth thirty minutes of your time, although my eyes glazed over at the encoding/decoding diagrams. I don’t have any background in it, but if you are versed, I’m sure you’ll get it.
Overall, this is a fascinating technology. It does actually make sense that the repeated re-attempts to send packets down the line would slow down the Internet traffic, and obviously, any solution to speed that up is good.
Added pluses to the technology include the fact that network coding enables a smarter node.
This is because, instead of data always traveling the same path, as is the case with TCP/IP, with network coding it can be routed over multiple, always-differing paths, or multipath, and thus is more secure. Dario Borghino mentioned this in an in-depth Gizmag article on the subject.
What is a little difficult to grasp is what patented, arithmetical equations are being used and why they work. However, Fitzek does promise to provide training courses in the technologies, and Steinwurf does have documentation up at its site and GitHub now, if you want to look further.
Naysayers are obviously not lacking, and Borghino’s article includes a fair amount of negative commenters. They include Christopher, who says that Internet speed is restricted by switch power, not packet paths; and Ivan4, who says it’s all down to the quality of the connection anyway.
And, if you’ll excuse me, my Netflix movie has been buffering, and is now ready for me to watch.
This article is published as part of the IDG Contributor Network. Want to Join?