Ordinarily, we conduct wireless throughput testing in our West Coast offices, intentionally subjecting wireless products to the crowded radio frequency environment that is downtown San Francisco.
In testing the Linksys Wireless-N Broadband Router and Wireless-N Notebook Adapter, we found that this harsh environment severely limited our ability to get a true measure of what the new technology could do because it tends to back off from full performance when legacy networks are detected. So, instead, we conducted our tests in the relatively clear airspace of a home office.
During testing, we used a variety of laptops from Dell (Inspiron 600m and Latitude D600), Hewlett-Packard (HP Compaq NC6320) and Lenovo (ThinkPad T41). The HP and Lenovo laptops played the part of our wired clients and servers; we installed the wireless adapters and drivers in both Dell laptops.
To measure bandwidth, we used the iPerf 1.7.0 client/server-architected benchmarking tool, which is freely available from the Distributed Application Support Team of the National Laboratory for Applied Network Research.
The routers we tested (based on Version 1.0 of the 802.11n draft standard and on Airgos Gen3 True MIMO chip set) each have a 10/100M-bps Fast Ethernet switch. The wired side of the network therefore becomes the potential bottleneck, as the wireless network—at least theoretically—should be able to best 100M bps. To test the full capacity of the wireless network, we had to employ bidirectional traffic to use the full-duplex capabilities of the switches.
In our maximum throughput test, we configured a laptop with the Linksys wireless card as the iPerf server. (The laptop was located 5 feet from the router.) We configured our two wired computers as iPerf clients, setting up the test so that one iPerf client would send data to the wireless-enabled iPerf server while the other machine simultaneously received data from the wireless client.
In our distance test, we measured the maximum upload throughput we could generate via a single wireless-enabled laptop sending data to one iPerf server connected via a cable to the router.
We tested performance at distances of 5, 25, 50, 87 and 102 feet between the wireless-enabled laptop and the wireless router. The 5-foot and 25-foot distances were on the same floor as the router under test, with no walls intervening. The 50-foot distance included two flights of stairs and two walls in between the laptop and the router. The two farthest distances included a third wall. (At this point, we and the laptop were outdoors and, for the farthest distance, across the street from the router.)
For the upload/download tests , we used the same testbed configuration as with the maximum throughput test. However, in this case, we configured both wired clients to simultaneously transmit in the same direction.
For each test, the results were based on the average of three successive test runs.
To find the clearest channel for our test, we used AirMagnet Survey 2.0 to identify any potentially strong sources of interference in the test area. We also used AirMagnets Spectrum Analyzer 2.0 to monitor the spectrum during tests to spot any sudden changes in the environment.