Mobile Hotspots: Which Are Fastest, Most Reliable?
No one wants to have to purchase separate wireless coverage for every device--wired or mobile--they do their computing on. Consumers want to buy a single connection to the wireless carrier’s network, one that will travel with them and connect any of their devices as needed.
As wide-area networks (namely 3G, 4G) get faster and better at reaching inside buildings, users might be able to cancel their DSL or cable service and rely wholly on one multipurpose wireless connection. We’re not quite there yet, but consumers can at least use a mobile hotspot to supply Internet access to multiple mobile devices, including laptops, tablets, and e-readers.
A mobile hotspot connects to the Internet via a cellular network, and then creates a Wi-Fi hotspot that can connect any Wi-Fi-enabled device within about 30 feet.
Mobile hotspots come two forms: as free-standing devices like Novatel’s popular MiFi, or built into smartphones such as Samsung’s Epic 4G or Motorola’s Droid X. We wanted to know which sort of hotspot--phone-based or free-standing--performed the best in the wild.
For three of the four major wireless carriers, we assembled a phone-based mobile hotspot and a free-standing mobile hotspot that each connected to the wireless networks of AT&T, Sprint, and Verizon. We used each device to provide Wi-Fi Internet access to both a laptop (a recent MacBook Pro) and a tablet PC (an Apple iPad). The one exception was the iPhone 4, which cannot act as a mobile hotspot but can tether to a laptop using Bluetooth or USB cable.
We went on a drive test, visiting six San Francisco locations and testing each viable combination of hotspot and laptop/tablet. We conducted three performance tests for each combination at each location using the FCC-approved Ookla mobile speed test, for a total of 18 tests per hotspot-device combination. We then averaged the results of those 18 download-speed tests for each device pairing.
Please click on the thumbnail to the left to view the full results.
On average, the Verizon MiFi delivered a download speed of about 1.5 megabits per second and an upload speed of 0.5 mbps both when connected via Wi-Fi and with a USB tether. The MiFi also delivered about the same connection speeds to the MacBook Pro as it did to the iPad. In general, the hotspots we tested could not connect a tablet as fast as they could a laptop. But the MiFi produced a respectable 1.5-mbps connection for the Apple iPad.
How does Verizon’s phone-based hotspot compare? Not very well. The Motorola Droid X connected the MacBook Pro at download and upload speeds far slower than those of the MiFi, and although the Droid X connected the iPad a bit faster, the speeds (download 1.3 mbps, upload 0.4 mbps) were still not as good as the MiFi’s.
AT&T currently doesn't sell a mobile hotspot (in fact, it seems to dislike the concept altogether), but you can insert an AT&T USB modem into a third-party hotspot--in our case a Cradlepoint PHS300 hotspot device--and achieve the same effect. And that effect proved to be pretty impressive in our trials. The Cradlepoint/AT&T USB pairing created the fastest Wi-Fi connections of any hotspot in our tests, especially when we hooked up the MacBook Pro. The combo seemed to grab the AT&T 3G signal and effectively turn it into Wi-Fi coverage while not sacrificing much speed at all.
For the laptop, the Cradlepoint/AT&T USB combination delivered download speeds twice as fast as those of the MiFi (albeit on a different network) and more than three times as fast as the Droid X hotspot.
The Cradlepoint also connected our iPad faster than any other hotspot did, at about 1.7 mbps on average.
Like the Cradlepoint device, the iPhone 4 didn’t seem to throttle the speed of the AT&T HSPA network much at all. The iPhone 4 lacks the Wi-Fi hotspot capability of the Droid X and the Epic 4G, but it can connect a laptop via a USB cable or Bluetooth. Although it didn't hit the download speeds that the Cradlepoint produced, our tethered iPhone 4 connected the MacBook Pro at a considerable 2.6 mbps for downloads, and it performed slightly better than the Cradlepoint on upload speeds.
We had two potential 4G devices with us, the Samsung Epic 4G and the Sierra Wireless Overdrive, both on Sprint’s 3G/4G network. Despite rumors of some Sprint 4G coverage being available in San Francisco, we couldn't find it--so our testing did not include the potential upside of 4G service for these devices.
Neither the Overdrive hotspot nor the Epic 4G produced blazing speeds in our tests. More important, neither device proved clearly better at delivering fast 3G speeds to our connected devices. The results, however, may be very different when a 4G signal is in the air. Our earlier tests in 4G markets showed that the Overdrive could connect to 4G at much higher speeds--and more reliably--than the HTC EVO 4G phone we were using at the time.
You can expect substantial performance increases for all these networks over time as AT&T moves to HSPA+ (by the end of 2010) and then to LTE, as Sprint continues to roll out WiMax, and as Verizon begins to light up its LTE footprint (in the initial 30-odd cities by the end of 2010). HSPA+ and LTE hotspot devices are not yet on the market, though, and even Sprint’s 4G service is not commonly available.
Phone-Based Hotspots Not as Reliable
In addition to speed performance, we scored each hotspot device on its reliability--that is, its ability to make a usable Internet connection. We ranked our test devices on a scale of 1 to 5, with 1 meaning "very unreliable" and 5 meaning "highly reliable." We based these scores on the number of times a given device restarted itself unexpectedly (or required rebooting), interrupting the data connection to all the devices linked to it.
All of the hotspot devices we tested connected in every testing location--some better than others. It is very annoying to see lots of "bars" of service and have a network connection, yet be unable to do any useful work.
The MiFi 2200 and the iPhone 4 were flawless, seamlessly connecting devices to the Internet.
Regrettably, the Droid X hotspot and, to a lesser extent, the Epic 4G hotspot weren’t so easy to use. These devices often reset themselves or just stopped working, requiring a manual reset. Sprint’s Overdrive often powered itself down and then up during our tests, rendering the device almost unusable for longer sessions.
Enabling/disabling the hotspot often solved those problems, but the less-than-stellar performance of these hotspots did little to make us forget the inconvenience. Even more annoying than having to manually reset the hotspot is to wait for a hotspot that resets itself.
After all the testing was done, we couldn't escape the conclusion that there is no perfect hotspot solution. Free-standing devices like the Novatel MiFi and the Cradlepoint PHS300 seem to perform better than phone-based hotspots (the tethered iPhone notwithstanding), delivering both better speeds and better reliability.
However, free-standing devices require us to carry around yet another piece of gear in our bag. Ideally, our smartphones, which we need for all kinds of other reasons, would also act as our mobile hotspot. Sadly, the smartphones in our tests offered less-than-impressive connection speeds, and often couldn't be relied upon to create an uninterrupted Wi-fi signal for the devices depending on it.
The speediest performer in our tests, the Cradlepoint PHS300, has no connectivity of its own, but relies on an AT&T modem card to connect to a network. However, there is word that Cradlepoint will soon be releasing a true hotspot device that makes its own connection to the WAN--no USB stick required. But for now, of the stand-alone mobile hotspots we tested, we give the nod to the Verizon MiFi for having the best combination of speed and reliability.
Stay tuned. As new 4G networks begin lighting up more and more markets, we're likely to see much better performance from both phone-based and free-standing mobile hotspots. The whole "mobile hotspot" product category is, after all, still young; as time goes on the software inside mobile hotspots themselves will surely get smarter, faster and easier to use.
PCWorld assistant editor Patrick Miller contributed to this report.