It's a safe bet that if you haven't jumped on the wireless bandwagon yet, you will very soon. Wireless networking is taking the world by storm, but the technology is changing at a quick clip, and new products and specifications are arriving faster than you can say, "Mr. Watson, come here. I want you." Our tests are designed to provide an objective analysis of the latest networking gear. To deliver this, we rely on industry-accepted network analysis tools, such as IXIA's IxChariot test suite. Our thorough testing emulates the real-world user experience, delivering key data and analyses on maximum throughput and range capabilities.

Test environment

We test the throughput and the range of PC Card network adapters in a two-node network environment consisting of a 1.4GHz Pentium M-based Dell Inspiron 5000M notebook with the PC Card installed and a 3.4GHz Pentium 4 550-based desktop system connected to an access point or a router. PCI network adapters are tested in a two-node, ad hoc network environment with the PCI card in the desktop and with a PC Card adapter in the Dell notebook. USB network adapters are tested in the same environment as PC card adapters, with the USB network adapter installed in the Dell notebook, connecting to an access point or a router attached to the desktop system. All systems run Windows XP. We typically test wireless network adapters with an access point or a router from the same manufacturer. The notebook client acts as the IxChariot console, which is used to execute and monitor tests that are run between endpoint pairs.

Before we conduct our performance tests, we analyze the local RF spectrum to confirm that the environment is conducive to wireless testing and to identify the ideal channels to use during testing. We perform this analysis using an AirMagnet Spectrum Analyzer. The channels are selected by finding the channels with the best available combination of minimal interference, minimal number of devices, and least amount of utilization. Each of these three criteria also must meet minimal thresholds--we will not conduct testing if the thresholds are exceeded. This ensures that every networking device is tested under ideal conditions.

For information on setting up your own endpoint pair and running a test yourself, see our guide to testing your own network.

Configuration

For our throughput testing, we place the test systems 10 feet apart from each other, with an unobstructed line of sight. For our range testing, the test systems are placed 200 feet apart, in an office environment with walls. Because 802.11b, 802.11g, 802.11n, HomeRF, and Bluetooth adapters, not to mention appliances such as microwave ovens and cordless phones, all operate in the same 2.4GHz spectrum (although some operate in the 5GHz spectrum), we test in an area free of any RF interference. Under these conditions, we achieve maximum signal strength, so our results reflect a best-possible case scenario. As signal strength fades, many devices negotiate with the clients to transmit at slower speeds. Throughput speeds will decrease as range increases.

IXIA's IxChariot

CNET Labs uses IXIA's IxChariot and Performance Endpoints test suites. The Performance Endpoints software is installed on each client to execute the transactions sent by the test script and measure the performance metrics for test throughput, transaction rate, response time, and lost data. We use IxChariot's standard throughput and response-time scripts for general testing, and we run the tests using TCP and UDP. Additionally, we analyze exceptionally poor or anomalous performance of any network adapter by using the throughput.scr script included in the IxChariot test suite, which allow us to pinpoint the source of problems.

We use IxChariot's streaming scripts to analyze devices with enhanced quality-of-service (QoS) capabilities. In such cases, the technology determines the type of supplementary testing we perform; occasionally we modify a test script to reveal the peak performance of a particular product. Any test-script alterations we make are noted in the performance charts accompanying the review.

Range and throughput tests

CNET Labs performs its range and throughput tests in a real-world office environment; therefore, our tests do not represent range and throughput capabilities across all environments. You may get better results in an open-air environment, a spacious facility without walls, or the outdoors. An area such as ours, divided by numerous walls, leads to decreased range and throughput capabilities. Our tests, therefore, provide a benchmark for estimating the range and throughput of wireless networking devices in an indoor setting, with a relatively limited number of RF obstacles.

We set the wireless adapter's transmit power to its highest setting and its connection speed to its maximum. For a typical 802.11b or 802.11g card, these settings are generally 30mW to 60mW transmit power and fixed 11Mbps or 54Mbps throughput, although some cards may offer greater transmit power. Adapters based on other technologies or standards, such as 802.11n, may be capable of faster connection speeds.

For our maximum range tests, we run the IxChariot throughput.scr test script with the network adapter and the access point or router separated by a distance of 200 feet for 802.11b, 802.11g, and 802.11n devices. If a device fails to complete a test at 200 feet, we continue to reduce the distance by 25-foot intervals until the device can complete the test.

For our maximum throughput tests, we run the IxChariot throughput.scr test script with the network adapter and access point or router separated by a distance of 10 feet for 802.11b, 802.11g, and 802.11n devices. We run three iterations of this script and report the highest of the three scores.

We also run a maximum throughput test at 10-foot distance in a mixed-mode environment. This consists of running the IxChariot throughput.scr test with two connected systems, one connected in 802.11b and the other connected in 802.11g. For 802.11n products, we run one system connected in 802.11g and the other in 802.11n. We run three iterations of this script and report the highest of the three scores.

Performance results for all tests are reported in megabits per second (Mbps).