Wireless Networking: Faster! Farther!

Test Results

If your wireless network doesn't have to reach very far and if you value speed, you can save money--and get a USB print server in the bargain--by buying the U.S. Robotics combo. It was the top performer in our close- and medium-range tests, recording average download and upload throughputs of 33.1 mbps and 27.8 mbps, respectively.

In our close-range tests, we placed the router and PC Card about 6 feet apart in the same room. The Airgo-based products performed the worst over short distances: Throughput from the Netgear Pre-N line dipped as low as 15.9 mbps. Our results suggest that MIMO products tend not to provide much of a performance boost within such proximity. In our most recent previous Wi-Fi roundup--one that did not include any MIMO products--a Buffalo Technology router achieved an average throughput of 27.6 mbps at close range, which is only a little slower than the 33.1 mbps achieved by the U.S. Robotics Wireless MaxG.

Because many users will be combining MIMO gear with older equipment, we ran a second set of close-range tests with a standard 802.11g client attached to the network. The performance of the Belkin, D-Link, and Linksys products remained essentially the same with and without the g card on the network. The other setups took only a slight speed hit; for example, the average throughput for the U.S. Robotics router declined from 33.1 mbps to 28.1 mbps. In this respect, they differ markedly from previous MIMO products, which incurred significant declines in performance when devices lacking their proprietary enhancements were added to the mix.

We didn't test speed and range for a standard 802.11g card on a network with a MIMO router, so we don't know how fast and how far a notebook with standard Wi-Fi would operate on one of these networks. But in April's "Stretching Wi-Fi" story , we discovered that the range for the standard 802.11g cards improved significantly. This result suggests that simply upgrading your router might deliver the performance boost you need.

In our midrange tests, we moved the notebook to a bedroom located two rooms and several household obstacles (or about 30 feet) away from the router. Performance was only marginally slower than at close range. Again, the U.S. Robotics setup emerged on top; and the three Airgo-based lines turned in the worst numbers, with the Netgear Pre-N products finishing in last place at 15.1 mbps.

Performance results were more or less reversed, however, when we took the notebook outside and tried to connect from a far corner of the backyard--about 100 feet away from the home office--for long-range testing. Here, the previously unimpressive Netgear Pre-N combo achieved the top average throughput, 15.3 mbps, followed closely by the other two Airgo-based product lines (from Belkin and Linksys). The U.S. Robotics gear, on the other hand, failed to complete the long-range test. Considering these results, if range and coverage are your top priorities, you'll likely be happy with an Airgo-based unit.

You might wonder why our test results fall far short of the theoretical maximum of 108 mbps for MIMO products. One reason is that performing Wi-Fi testing is notoriously tricky: Radio waves are subject to all sorts of interference such as from physical objects, other radio waves, and even the weather. Equipment placement is another speed variable. The greater the distance between a Wi-Fi router and a network card, the greater the speed boost over standard 802.11g becomes--especially as standard g signals peter out at the outer edge of a network. Finally, vendors generally don't secure their networks when they test--because doing so can reduce performance--but we tested using WPA, the security technology we recommend to readers (see "Wi-Fi Security Options").

Routers that use Atheros Super G chips--among them, Netgear's RangeMax and D-Link's Super G MIMO lines--employ a somewhat controversial technology called channel bonding to boost the range and speed of their products.

Channel bonding works like this: In the United States, standard 802.11b and g equipment, as dictated by the IEEE standard, transmits data on one of eleven channels in the vicinity of the 2.4-GHz frequency. Of the eleven channels, only channels 1, 6, and 11 are spaced sufficiently far apart that all three can be used simultaneously without interfering with each other. The availability of these noninterfering channels enables standard 802.11x wireless equipment to deal with interference from nearby wireless networks; thus, if your network is using channel 1, your neighbor's equipment might automatically hop over to channel 6 or 11.

But equipment that uses channel-bonding technology beefs up throughput by hogging all three noninterfering channels, effectively shutting out all other networks within range when transmitting data; as a result, a neighboring network could slow down. The Wi-Fi Alliance considers the problem so significant that it will not certify products that don't, by default, stop using channel bonding when another network is in range.

We tested all of the wireless products for this review at their default settings, since that's the way most people will use them. Dismayingly, our Netgear RangeMax router continued to use channel-bonding technology by default--whether or not a neighboring network was in range--unless we added a network client that didn't use channel bonding. In other words, the RangeMax network wouldn't automatically step down from channel-bonding mode if a neighbor's network was in range, which we consider bad-neighbor technology.

Netgear representatives say that, since we received our evaluation unit, the company has altered the default settings on RangeMax routers to abandon channel bonding when the device detects another network nearby. Clearly, however, some people will buy products that shipped with the unfriendly setting as the default, and we docked a few points in our ratings for this reason.