We used a range of tests on the wireless equipment to gauge their abilities to perform under duress, as well as provide empirical evaluations on the set-up, configuration, management and software capabilities.
For the performance tests, we matched up the access points with the PCMCIA cards and when not possible, we tested it with the card of the same network type. Non-router access points were tested with a D-Link 804V router. As interoperability is a big part of the open 802.11 standards, we cross-pollinated hardware to gauge their ability to cope with other manufacturers' hardware and firmware.
Before the testing began, we took detailed notes on the user experience of actually setting up a network with the reviewed equipment. You'd be surprised at the meagre resources, manuals and user guides that most of these products came with. Unpatched software, dodgy firmware and out-of-date drivers are de rigueur it seems with wireless technology, so it's recommended you download all the files you need before you start.
Configuring and setting up the network is one thing, but as PC Authority has mentioned many times in the past – wireless network security is paramount, but often overlooked. In our tests we configured the networks with WEP (Wired Equivalent Privacy), using maximum settings for each device. We looked at WEP not only for operation, but ease of setup.
We used a PC as our static test bed, and an IBM ThinkPad R40 as our roaming notebook. All the tests were conducted indoors in a typical office environment. We chose two points in the office to perform identical tests – one within meters of our test bed where reception was perfect, and the second from a point 15 meters away but blocked by two walls, electrical cables and PCs. This second point represented one of the worst positions in the office to get reception, but without quite dropping out altogether.
We used three tests and performed them from each position three times: pings, transfers and streaming. We pinged the test PC 100 times and recorded the average response time in milliseconds, plus the percentage of dropped packets. This shows us the signal quality from both locations.
Our second test was to stream a movie file from one machine to the other. This was more a visual representation of the data-link quality as streaming video is a high bandwidth operation. Our third test was to transfer a 51.26MB file between the machines three times at each location, and to time how long these transfers took.
Although we garnered a lot of data on each piece of hardware, we stuck with only the most telling and relevant results for the graphs (above). These are the data transfer speeds you'll expect to see in perfect and dismal network conditions. Where hardware was tested on various manufacturers' products, an aggregate score was used. All speeds for all products are averaged.
The clear curves in the graphs show the speeds offered by the competing standards, and it's these speeds, more than the differences in hardware, that show the performance gains to be made from one standard to the other. A high score does not necessarily indicate a better product, although it may within standards.
TECHNICAL EDITOR
Darren Ellis
CONTRIBUTORS
Dan Chiappini,
Nirmal Chandrasena