We used two platforms. One test platform was the Fluke Networks AirCheck WiFi tester. The AirCheck contains a 2 x 2 antenna system, and thus can support only two spatial streams.
The second test platform was a Lenovo T-410 Laptop that was able to take advantage of the performance boost offered by a WiFi router with a 3 x 3 antenna configuration (see Joanie Wexler's article on how these configurations differ).
Each of the routers was placed in the same location at the top of an equipment rack in the lab. The device was placed on a defunct Intel Fast Ethernet switch, which would be a typical location in an office environment. Each device was located about six feet above floor level. Where necessary we changed the location and orientation of each device to get the best results possible.
We conducted several tests using the Lenovo laptop computer. One was to measure the distance over which the WiFi router could deliver a standard music video without exhibiting any data loss as evidenced by audio or video breakup, or pixilization.
The distance measurements took place on a surveyed test range that allowed measurements up to 300 feet. Testing required clear weather both because of potential damage to the test equipment if it got wet, and because of potential signal attenuation from rain or from evergreen trees. (When wet, evergreen needles can act like thousands of antennas, and could produce unpredictable results.)
FTP testing took place in the test facility at distances of 15 feet and 50 feet. The longer range included two exterior and two interior walls. We used the given transfer times as relative indicators of file transfer rates, and we ran them several times to find an average because the FTP transfer times varied widely. The FTP target was an HP xw series workstation with two dual core Xeon processors, directly connected to the network core switch by Gigabit Ethernet. The WiFi router was also connected to the same Gigabit Ethernet switch to reduce latency.
The performance tests conducted by the Fluke AirCheck consisted of a stream of 64-byte ping packets. The rate of packet transmission reflects the overall performance of the WiFi device. The AirCheck displays the actual bandwidth of the WiFi device as well as a chart showing the performance over time. The level of packet loss is displayed on the same screen. The test consisted of moving the AirCheck progressively farther from the WiFi device until the first packet loss appeared. We tested this through several iterations.
The streaming video test used Windows Media Player to stream video from a Buffalo LinkServer, also connected to the core switch using Gigabit Ethernet. Distance measurements were at the first indication of signal disruption, usually a breakup of the audio or video stream as the laptop was moved progressively farther from the WiFi device. Tests for signal loss included changing the orientation of the laptop antennas during several iterations of the test.
For Internet connectivity, we used a cable connection provided by Cox Communications that has a nominal speed of 25Mbps for downloads and 3Mbps for uploads, although Cox does not throttle its network so those speeds can be much higher in off-peak hours. This was reflected in all of the router test results when we accessed content from the Internet - it didn't matter how fast the router was, the speed of the Internet connection is the controlling factor.
We conducted usability testing by first following the instructions in whatever served as documentation delivered with the device. Usually this documentation was a poster or short pamphlet. On the first try, even if we knew that a step was being skipped in the instructions, we followed the instructions to determine if an inexperienced user would be able to configure the device. When that didn't work (which it usually didn't), we would start again and perform the obviously missing steps. In each case we first started with whatever the vendor delivered as a setup wizard, and performed a manual installation only if the wizard failed.
Once the initial setup was complete, we would then configure the device manually to set up security, network addressing, attached devices and the like, using whatever context sensitive help that was available, and using the manual where necessary. Except where noted, the setup wizards only performed very basic setup, and usually that didn't include any security. Usability testing also included inspection of the exterior indicators, markings that would help us understand or set up the device if the instructions were lost, or notices as to proper operation or diagnosis.