Wireless equipment vendors pushing a kit built to the draft 802.11n specifications will have to explain to customers why it apparently fails to deliver one of the main benefits of the faster Wi-Fi technology - better performance at a distance.

The 802.11n specification will not be complete until next year, but vendors have been rushing to be first to market with products based on an early draft. Analysts have warned that these products will perform less well than existing equipment based on more mature implementations the MIMO technology which will be used in 802.11n.

Now testing by Tim Higgins of Tom's Networking has shown that the products fail to deliver on a major MIMO promise - better performance at a distance. Draft-N products in fact perform worse than bog-standard 802.11g systems, as the distance from the base station increases. The testing goes further than earlier tests and commentary by Craig Mathias of Farpoint Group, and produced similar results.

Higgins tested products from Buffalo and Linksys, both based on Broadcom's draft-N silicon, and found that their "both... products throughput fall off earlier than the Airgo [the existing Rangemax240 from Netgear] and fail to transition to lower link rates so that they can stay connected." Netgear's Next Draft-N product, based on Marvell silicon did a bit better: it managed to shift down to lower rates, "but at a throughput lower than that provided by the 11g pair".

In other words, the Draft N products only outperform basic 802.11g Wi-Fi equipment at short ranges, where an Ethernet cable would be even better - and this indeed is the situation where we have seen Buffalo's Draft-N products demonstrated.

With this level of performance, Higgins describes the Draft-N products as "science experiments" - and his test might also be described as an experiment. To some extent it was also an examination of the test methodology, as Higgins used a set-up from Wi-Fi test specialist Azimuth Systems, which launched in 2003, and isolates the equipment from the "real" environment, replacing antennas with a manageable wire-based model of radio attenuation in a building.

Wi-Fi testing is controversial, and MIMO adds new issues. Since MIMO's effects rely on multiple paths set up in a real environment, Higgins did not test the multipath benefits of Draft N (Azimuth has a feature that does this, but Higgins kept the test simple). While vendors will argue that this invalidates the test, others are not so sure: "the fact that the Airgo-based device performed substantially better than the Draft N equipment would seem to defuse that argument," comments Glenn Fleishman in the MIMO + N department of his Wi-Fi Net News blog family.

“With high prices, immature drivers and firmware, no guarantee of upgrade to standard 11n when it’s released and now - shown for the first time - evidence that some current products don’t even perform better than 802.11g at lower signal levels," concludes Higgins, "I can’t think of an upside that justifies the expense and hassle.”

There's a sting in the tail for Netgear, already struggling to justify its launch of Draft N products that perform worse than its existing ones. Higgins reiterates an earlier result that the Netgear RangeMax 240 router based on Airgo's third generation chip also has the "bad neighbour" problem of interfering with existing 802.11g networks.