Skip to Content

How Apple Could Boost Speeds 20 Times on the Next iPhone

The new iPhone breaks ground by seamlessly sharing Wi-Fi and 4G for Siri. Further tweaks could boost bandwidth 20-fold in some cases.
September 30, 2013

A wireless networking technology found in Apple’s new operating system could—if tweaked—provide a 10- to 20-fold bandwidth increase in some situations, like on a moving train or in a busy urban environment, new research suggests.

The technology is called multipath TCP. It allows you to use multiple wireless networks—such as 4G and Wi-Fi—at the same time. But Apple isn’t using it fully, nor is it using an advanced version—one that also encodes the data being transmitted in new ways— recently shown to provide those dramatic potential gains.

The advance, based on work done by a multi-university group led by Muriel Medard, an electrical engineering professor at MIT, is “very compelling” and “shows dramatic improvement in terms of increased data rates, reduced latency, and reduced packet loss,” says Andrea Goldsmith, professor of electrical engineering at Stanford and a leading network researcher and entrepreneur not involved in the work.

Right now, as any smartphone owner knows, a phone or tablet will either use Wi-Fi or 4G or 3G—and never at the same time. So your streaming video may cut out because the network you were using dropped, even though there’s another signal available.

Multipath TCP could change this by divvying up those video bits across two or more networks. “Multipath” refers to using more than one wireless route, and TCP refers to the protocols used by most Internet traffic. Then, to use a simplified explanation—all “odd” packets (units of data that make up an Internet transmission) get sent over Wi-Fi and “even” ones over 4G. Then these “odd” and “even” packets get woven back, zipper-like, on the phone.

But in practice, it’s not that simple. The problems start with the fact that data-transmission takes longer from a cell tower than it does from a Wi-Fi router. Throw satellite streams in and the transmission delays are even longer.

Multipath TCP makes up for this by tweaking transmission speeds. But matters get more complicated if you are moving around, meaning those timings are always changing—and worse still, if some packets drop out. When those things happen, the computation required for multipath processing can get so complex that it actually slows down the overall speeds, says Medard.

And that may be why Apple—in using multipath TCP only for its voice-query engine, Siri—apparently isn’t even using both cellular and Wi-Fi networks at the same time. Rather, it may be using the technology to simply enable Siri to switch back and forth between them without user intervention, so it can avoid having to retransmit your spoken request, a source of delays.

“The rumors I’ve heard is that Apple is using it for Siri just to decrease latency by using whatever network connection is available,” says Jason Cloud, a grad student in Medard’s lab.

Trudy Miller, an Apple spokeswoman, declined to comment on how the technology has been deployed. (Apple has been characteristically secretive about the technology; so one of the first hints that it was using multipath at all came when a Belgian researcher, Olivier Bonaventure, blogged about it earlier this month.) Several groups around the world are working on the technology.

At any rate, there is a technology that helps solve the remaining problems with multipath TCP. It is called “network coding”—an extra tweak atop multipath TCP. Network coding algorithmically combines packets in elegant ways. Then multiple packets can be turned into a single number that’s a function of the ones making it up. “You code within flows for redundancy,” Medard says. “Then you don’t have to be managing between them like crazy.”

It was this version that, when tested by Cloud and colleagues at the Hamilton Institute, part of the National University of Ireland in Maynooth, Ireland, provided up to 10 times better performance on a single network path, by repairing dropped packets on a single connection. This expanded on work Medard did last year (see “A Bandwidth Breakthrough”).

Measurements done at the University of California, Los Angeles, suggested how network coding could turbocharge multipath TCP. In findings presented in June, Medard and several university collaborators measured actual packet losses and other conditions around a Westwood, California, campus from three wireless sources: Wi-Fi transmitters, cellular towers, and Iridium satellites. They concluded the technology could provide a similar benefit when used on multiple paths, with a potential tenfold increase per path.

Demonstrations are planned over the next year. But commercialization of the underlying network coding technology is already possible. Medard says several organizations have licensed the technology from an MIT-Caltech startup called Code-On Technologies. She says she can’t name the companies.

Goldsmith says that while she didn’t know the details of what Apple has done, “it’s great to see Apple move in this direction, as it will inspire more developments in theory and practice.”

While “it is a no-brainer” to move forward with industry adoption of network coding with multipath TCP, implementation will likely have to start with applications where the data can easily be coded and decoded at either end—like, say, on a video application. “It’s very difficult to fundamentally change the network. There are a lot of entrenched players and entrenched technology,” Goldsmith says.

Keep Reading

Most Popular

Large language models can do jaw-dropping things. But nobody knows exactly why.

And that's a problem. Figuring it out is one of the biggest scientific puzzles of our time and a crucial step towards controlling more powerful future models.

How scientists traced a mysterious covid case back to six toilets

When wastewater surveillance turns into a hunt for a single infected individual, the ethics get tricky.

The problem with plug-in hybrids? Their drivers.

Plug-in hybrids are often sold as a transition to EVs, but new data from Europe shows we’re still underestimating the emissions they produce.

It’s time to retire the term “user”

The proliferation of AI means we need a new word.

Stay connected

Illustration by Rose Wong

Get the latest updates from
MIT Technology Review

Discover special offers, top stories, upcoming events, and more.

Thank you for submitting your email!

Explore more newsletters

It looks like something went wrong.

We’re having trouble saving your preferences. Try refreshing this page and updating them one more time. If you continue to get this message, reach out to us at customer-service@technologyreview.com with a list of newsletters you’d like to receive.