Wireless LANs are a natural fit in the home and, for some, a constant source of fits. Standard 802.11 WLANs suffer limited throughput, which only worsens as the distance between the client and the access point/router increases. The highly variable nature of radio propagation means seemingly short distances can yield poor results, especially when walls and floors get in the way.

To improve conditions and make WLANs robust enough to stream media, hardware vendors have begun applying multiple-input/multiple-output (MIMO) technology. MIMO uses multiple antennas on each end of a radio link to send and receive several unique radio signals in a single radio frequency channel. The ability to send multiple, distinct datastreams over the same signal simultaneously, known as spatial multiplexing, can double and even triple data throughput rates (though vendors are arguing over the definition of MIMO).

Moreover, sophisticated algorithms create a signal that's louder, which translates directly into corresponding improvements in performance.

Belkin entered the consumer market last fall with the first MIMO-based products, built using Airgo's chipset. These products, called pre-N, in reference to the IEEE's upcoming 802.11n standard, are reviewed here.

Since last fall, Cisco's Linksys subsidiary, D-Link and Netgear have all shipped "MIMO-based" products. However, the technologies they use and call MIMO vary considerably. While all three use multiple antennas to send and receive multiple datastreams over the same signal, only the Linksys SRX line uses spatial multiplexing (like Belkin) to transmit multiple, distinct datastreams over the same signal.

Don't expect compatibility with the final 802.11n standard
These first MIMO-based products won't be compatible with those based on the final 802.11n specification. However, they are fully Wi-Fi-compliant. Three important questions remain: Do today's MIMO-based products provide sufficient performance to justify the cost? How does the performance compare with standard 802.11g products? Do they cause interoperability problems on a mixed network?

To find out, we devised a series of tests to determine how MIMO-based products would fare against standard 802.11g gear. To mitigate the effect of radio-related artefacts, we rented a house for the sole purpose of running these tests. It was empty except for our equipment and us.

To benchmark performance, we used Iperf, a free LAN test suite. We ran the same test in each case; the only variables being the devices tested and their locations. Because location can't be reproduced precisely with antenna orientation, we placed the client notebooks on turntables revolving at 45 seconds per rotation. This let each radio cycle pass through a range of orientations, factoring out any overly beneficial (or detrimental) positioning. Two full cycles (90 seconds) defined each run.

We tested two MIMO products, Belkin's pre-N router (F5D8230-4) and PC card (F5D8010); and four standard 802.11g products - the Linksys WRT54GS router and WPC54GS PC Card, and the Netgear WGU624 router and WG511T PC Card. (We turned off the power management on our Dell Inspiron 8600 notebook and used default driver settings.)

The Iperf benchmark ran in three configurations. First, we put the client and router in the same room 13 feet apart, which served as a baseline. For the second, we had the client on the same floor as the router, 45.8 feet away and with three walls between them. For the third, we had the client upstairs, 46.7 feet away from the router, with four walls and a floor between them. We tested all combinations of clients and routers, and examined upstream and downstream performance.

Pre-N gives better throughput and range
Not surprisingly, the MIMO-based products from Belkin yielded the best throughput and range. In the homogeneous tests (client and router from the same vendor), the non-MIMO configurations had only 41 percent to 72 percent of the throughput of the pure MIMO-client and router. Moreover, a mixture of MIMO and non-MIMO products yielded better results in every case than a homogeneous, non-MIMO configuration. This shows that MIMO provides a benefit, even when implemented on only one end.

We experienced no interoperability problems - the MIMO gear worked well with all the other equipment. Finally, having MIMO on at least one end of the connection let us establish a link when we couldn't establish a homogeneous connection.

The pure Belkin configuration provided nearly 15 Mbit/s of throughput in the upstairs test, where the homogeneous configurations of the other products usually failed to connect at all.

Our interoperability tests put to rest any concerns about the ability of Belkin pre-N products to work with standard 802.11g offerings. The performance improvements with MIMO-based products on only one end of the connection were impressive.

We recommend these first MIMO-based products to residential users. There is significantly better throughput and range performance than conventional products in every case we tested. MIMO-based is a better approach than adding third-party, high-gain antennas or active repeaters - it's less complex and, even with the higher prices, usually less expensive.

Mathias is the principal analyst at Farpoint Group. This article appeared in Network World