Vendors like to show off high-profile case studies, but maybe an airport is a touch too public, if your rivals have a point to make?
At the end of 2007, wireless vendor Meru announced it was providing Wi-Fi at Denver Airport. It's not taken long before the company's rivals got round there, started poking, and announced that the installation wasn't doing what it claimed to do.
Meru has a distinctive Wi-Fi architecture, which puts all the access points on the same channel. As I've understood it, when a mobile device moves from one access point to another, roaming is managed from a central controller, and the mobile device appears to be on the same access point all the time. But rivals say that is not what is happening at Denver.
I've been sent a document, apparently created by an Aruba engineer who happened to be passing through Denver with an analyser. His scans reveal that all the access points are indeed set to one channel (channel 1, wihch is bonded to channel 6 to create the dual-width 40MHz channels which 802.11n fast Wi-Fi allows). They all showed the same ESSID, (diafreewifi for Denver Airport) but the access points each have a different BSSID (their MAC address). This means that the multiple access points are not hidden, and roaming is under the control of the mobile device, not the controller.
I put this to Rachna Ahlawat, vice president of marketing at Meru, at last week's Wireless 08 event in London. She agreed - "We have single channel at Denver, but not the full Virtual Cell," she said. Handover is indeed initiated by the client at the moment, she said, but that was part of Denver's requirement. The airport expects to move to Virtual Cell soon she says, but already gets significant benefits.
Looking back at coverage of Denver, I can't see anything that goes into enough detail to specify whether it would use Virtual Cell or not.