Making sense of 3G speeds
Reports abound that the iPhone 3G isn’t living up to its family name: it may be an iPhone, but ordinary owners, reporters, and pundits are saying that in regular usage its actual speed is far below the potential 700 Kbps to 1.7 Mbps claimed by AT&T for its currently deployed HSPA (high speed packet access) third-generation cellular data network.
I have a slight problem with this as someone who has obsessively followed first the evolution of Wi-Fi into something as nearly available as oxygen, and then cellular data networks as they moved from slow 2G into modest 3G rates.
Neither Apple nor AT&T has ever promised those rates to its iPhone subscribers. At the iPhone 3G’s introduction in June, Steve Jobs didn’t state the speed of the 2.5G EDGE standard that’s the fastest supported in the original iPhone—as fast as 200 Kbps with AT&T’s version—and then say that the iPhone 3G would be 3.5 to nearly 8 times faster. Instead, he showed a couple of examples, and talked about ranges of 2 to 3 times faster.
If you visit Apple’s Web site, you’ll note that there’s no mention of speed per se: the only mention right now is of a “2.4x” speed improvement (20 seconds instead of 48 seconds) for loading lonelyplanet.com. That may be disappointing, but it’s clear that Apple is underselling the speed.
AT&T is a bit broader, stating on its iPhone page that “iPhone 3G harnesses the power of AT&T’s broad and powerful 3G mobile broadband network, which offers 3G mobile phones download speeds of up to 1.4 Mbps,” with a footnote that doesn’t qualify the statement at all, noting a comparison that doesn’t exist with the original iPhone model. “Up to” are weasel words, but you’ll note that AT&T separately promises its LaptopConnect customers a typical performance range, with 700 Kbps at the low end. (You’ll see lower downstream rates on some AT&T mobile broadband pages than the 700 Kbps to 1.7 Mbps rate, but AT&T apparently doesn’t update its Web sites very consistently even after they make major press announcements on speed boosts.)
The iPhone is bound by a processor that has to fit its confines, not use too much power, and handle the dozens to hundreds of simultaneous tasks needed for a portable computer that’s also constantly connected to a cellular and/or Wi-Fi network. The iPhone 3G is just as limited by its CPU as its predecessor.
I don’t own an iPhone 3G, but in several weeks of testing of a loaner unit, I didn’t find much difference between using the iPhone on a 3G network and on a 3 Mbps-backed Wi-Fi network. The Safari browser can only load and render pages so fast.
PC World colleague Melissa J. Perenson wrote an excellent rundown of her experience in gauging the iPhone 3G speed last week, but she overlooks a few points.
First, AT&T and Apple don’t promise the rates for the iPhone that she cites for their 3G network, as I note above.
Second, the only way to tell whether AT&T’s 3G network is congested or at fault is to test an iPhone 3G side by side with a laptop using an AT&T LaptopConnect network card with the latest HSPA standards built in. Running speed tests on both the iPhone 3G and a laptop in the same place would reveal whether chips or networks are at fault.
Third, performing a laptop and iPhone test on a high-speed Internet network where the devices are connected by Wi-Fi would further show whether the iPhone 3G’s processor was limiting the page rendering speed more than a lack of downstream throughput. (When Steve Jobs said 3G is faster than Wi-Fi at the Worldwide Developer Conference, I had to laugh. Wi-Fi using 802.11g can provide more than 20 Mbps of throughput; it’s the backhaul to the Internet, the connection of the Wi-Fi network to the rest of the world, that’s the limit. 3G tops out at a peak rate of 1.7 Mbps on AT&T’s network, and backhaul problems from cellular towers—which might need to carry 10 Mbps or more back to the Net—are one of the most expensive and serious problems facing 3G network expansion and future 4G network build out.)
Fourth, AT&T only covers a percentage of the U.S. with its 3G service. It’s somewhere close to or north of 300 of the top 350 metropolitan markets. And AT&T’s coverage tends to be metro area—on the edges, in suburbs or even certain neighborhoods, you’re going to drop down to 2.5G for data and 2G for voice. This makes it hard to know whether those complaining about speed, even if they’re consulting a coverage map of their area, are in a marginal zone or not. In the heart of a city, you shouldn’t have trouble, except for the dirty little secret that concrete canyons can reflect and absorb and block signals, too. (There are similar complaints about the iPhone 3G elsewhere, too, of course, and those other carriers, too, only cover a percentage of their country with 3G.)
My biggest problem with the complaints about the iPhone 3G isn’t the disappointment people have. I’m not telling them (or you) to suck it up. People expect 3G to be much, much faster than 2.5G because of all the hype that’s been layered onto 3G. 3G networks in my testing over the last couple years work remarkably well, and I’ve been able to see every cell carriers’ top speeds (and their middling speeds, too) wherever I’ve tested. That surprised me. I used to be quite bearish on 3G, but seeing (and testing is believing).
No, rather, the problem is that we—the Mac and telecom media—have no way to know whether this problem is widespread or limiting to people unhappy enough to post on Apple, Macworld, and other forums. Unless this fine publication were to commission a study to get 1,000 iPhone 3G users selected at random to report their satisfaction with 3G speeds, and, more difficult, to provide statistics of their usage, I don’t see how anyone knows how widespread or severe iPhone 3G throughput issues really are. They could be terrible. They could be mostly terrific.
Wired is making an attempt to gather data, but it’s self-reported. Self-reported data is entirely unreliable unless you have another loop that lets you calibrate self-reporting information against a random sampling. (If you could get very precise information from 1,000 people and 100,000 others self-report, you might be able to correlate the two bodies of information to correct for self-reporting bias. Or not.)
In the end, how can we tell whether the iPhone 3G is performing as expected—or as people expect it to (which are two different notions)? Word of mouth. Message boards and reporting represent a subset of people willing to talk about or write about a given problem, and both suffer from the squeaky-wheel syndrome. But if the iPhone 3G sales fall below expectations, that might reveal that the word of mouth among actual users to their friends and colleagues drops interest.
After spending weeks with a loaner iPhone 3G, I don’t bad mouth it: the faster speed is quite nice, especially with audio streaming iPhone applications, and the GPS has worked extremely well wherever I’ve used it. But I’m not shedding my 2.5 iPhone in favor of an iPhone 3G, cost aside. There may or may not be a problem in the 3G phone’s innards, but I don’t find the speed bump a big enough one to get rid of my plain old EDGE iPhone.
[Macworld contributor Glenn Fleishman writes daily about wireless networking at his site Wi-Fi Networking News.]