802.11g WLAN gear

D-Link, Belkin top our early wireless tests

The lack of a final standard hasn't prevented an early crop of 802.11g wireless LAN products (54M bit/sec, 2.4-GHz frequency) from coming to market. We obtained seven prestandard access points and put them through their paces. This gear is pointed largely toward home and small business markets, and the product features show it. This first group is far faster than pure 802.11b, yet slows quite a bit when 802.11b clients try to access them. They show promise, but our results will change when a final standard brings compliant firmware upgrades.

To continue reading, register here and become an Insider. You'll get free access to premium content from CIO, Computerworld, CSO, InfoWorld, and Network World. See more Insider content or sign in.

The lack of a final standard hasn't prevented an early crop of 802.11g wireless LAN products (54M bit/sec, 2.4-GHz frequency) from coming to market. We obtained seven prestandard access points and put them through their paces. This gear is pointed largely toward home and small business markets, and the product features show it. This first group is far faster than pure 802.11b, yet slows quite a bit when 802.11b clients try to access them. They show promise, but our results will change when a final standard brings compliant firmware upgrades.

We tested products from Apple, Belkin, Buffalo Technology, D-Link Systems, Linksys, Netgear and SMC Networks. Buffalo's products hit our doors first - and, like many of the participants, the company followed up with several CDs or fat firmware files worth of revisions in the three months after products were released.

The D-Link DI-624 and Belkin's 54g products win our Blue Ribbon Award - Belkin's installation, management and performance were very good, and in terms of raw performance, D-Link's DI-624 was positively blazing - outdistancing the competition.

But are any of these products ready for production deployment? The answer is mixed. With the 802.11g specification still in draft form - it's unlikely that any of the products reviewed will have the same firmware after the final specification is agreed upon this summer. All these products are upgradeable, but it's difficult to tell how the changes will affect performance. At the time we conducted our test, the products lack ed support for the upcoming Wi-Fi Protected Access specification (see "Securing 802.11g"), so high security is up to third-party authentication schemes.

Speed and range

Because 802.11g uses the same radio spectrum as 802.11b and also provides optional backward-compatibility with 802.11b, 802.11g is perceived as the successor to 802.11b (See "The 'Frankenstein' spec").

The draft specification provides for speeds up to 54M bit/sec, but in testing we found on average about 14.97M bit/sec of speeds. The difference between the gross speed and our average results is similar to the differences seen in 802.11a and 802.11b products - signaling overhead eats up a large fraction of performance (roughly more than one-third off the top). The data rates with 802.11g can be blazing compared with 802.11b. D-Link turned in the best optimized speed, with a 24.73M bit/sec rate.

But these speeds often are compromised if an 802.11g access point is in dual-mode 802.11b/g - because the access point will fall back to the slower data rates if there's an active association with any 802.11b device. Some access points were more sensitive to fallback data rates and might be perceived as slower as a result. For optimum results (and for the basis of grading performance), we used the products' pure 802.11g modes where available.

When an 802.11b client was used (various 802.11b-only cards) in close proximity to the access point, data rates would drop back to 802.11b levels. Mixed mode performance drops are typical and part of the draft standard. This effect likely will continue to change until stabilized under the final standard when it is agreed on in the early summer. Because of the expected changes that will take place in the standard, speeds are likely to change when subsequent firmware arrives for each access point.

Geographic coverage was fairly uniform across the products when set in the strict 802.11g mode (where possible). D-Link's DI-624 had the largest range (344 feet), and Buffalo's gear had the smallest radio range (about 213 feet) before signals degraded to uselessness. In 802.11b/g mode, the range for all of the access points was smaller - on average about 279 feet. There was not much of a range difference between 802.11b and 802.11g modes, although our speed fell back slower in the pure 802.11g modes as we moved away from the access point.