eWEEK Labs latest tests of products based on the IEEE 802.11n draft standard measure not only the top-end performance of products based on the specification but also the level of interoperability among different vendors products and the performance degradation caused by encryption.
We used Ixias IxChariot 6.25 to measure performance, using Ixias High Performance Throughput test script modified to run for 15 seconds for each test run.
Three routers we tested included only Fast Ethernet switches, so each test mix contained an upstream and a downstream connection between the wireless client and the IxChariot server. With this full-duplex traffic, we could avoid a bottleneck on the wired side of the network and measure the real performance of the wireless side.
Each performance number published in this package is culled from the results of five concurrent test runs. We dropped the highest and lowest values and then took the average of the remaining three.
We performed the tests using a pair of Lenovo Groups ThinkPad T60 laptops—one acting as the wireless client and the other as the IxChariot server on the wired network.
For each client we tested, the driver and client supplicant application were installed on a clean, fully patched Windows XP operating system with Service Pack 2. We also installed Windows XPs WPA2 (Wi-Fi Protected Access 2) patch (KB893357) on each client machine.
We conducted our Unencrypted Range tests at four different distances. With the shortest distances (10 feet and 40 feet), the wireless router and client were in the same room with unobstructed line of sight between devices. At the 55-foot mark, the client was elevated two floors above the router, approximately 22 feet up with two walls in between. The furthest mark—95 feet—adds a third wall between devices.
For our encrypted tests, we configured the wireless link to use AES (Advanced Encryption Standard) between the client and router. Because we were concerned mainly with the performance hit on the fastest connection, we have shown only the values at 40 feet.
We also show the measurements at 40 feet for our interoperability matrix. This matrix consists of tests between each router/access point under test against every client adapter. This chart shows not only the interoperability level of the draft 802.11n products but also the performance in conjunction with legacy 802.11g products.
To gauge backward compatibility, we measured each products capabilities in conjunction with standard 802.11g gear. We used a Cisco Systems Aironet 1200 Series access point and the latest Intel Centrino adapter.
The Buffalo Technology and Linksys routers we tested support automatic channel selection, and we allowed these routers to define which channel was used during our tests. The Netgear and Belkin routers we tested do not offer automatic channel selection, and, for those products, we configured the 40MHz channel for Channel 7 after using NetStumbler.coms NetStumbler 4.0 to ensure that no other Wi-Fi devices were transmitting in the vicinity on nearby channels.
We monitored the RF (radio frequency) environment during tests using Cognios Spectrum Expert for Wi-Fi.
Technical Analyst Andrew Garcia can be reached at andrew_garcia@ziffdavis.com.
