Ruckus Wireless looking to cause a commotion with its critique of Cisco's Wi-Fi technology

From John Cox, Network World senior editor and blogger: The scrappy folks at aptly-named Ruckus Wireless are damning with faint praise Cisco's recent announcement of beamforming technology to its new 802.11n access point, the Aironet 1140. 

In our coverage of the 1140 announcement, we cited Ruckus' alternative approach -- using its patented multi-element antenna and software to select the optimal Wi-Fi signal from the AP to a given client. By contrast, Cisco is implementing the optional 802.11n beamforming in the Wi-Fi silicon, with the name ClientLink.

Despite being part of the 11n standard, the Cisco beamforming technology is actually targeted at improving not the performance fo 11n clients but that of "legacy" Wi-Fi clients -- 802.11a and 11g devices. Cisco cites independent test results showing some impressive gains for these devices linked to the 1140 access point.

In a press release today, Ruckus "applauded Cisco Systems for validating the value of beamforming in improving Wi-Fi performance..." and then went on to list its arguments about why the silicon-based model doesn't measure up to what Ruckus can deliver. 

Ruckus isn't alone. Devin Akin, the CTO at CWNP, an Atlanta, Georgia-based company focused on enterprise Wi-Fi training and certifcation for IT professionals, has a blog post that dissects some of the Cisco claims and documentation. Among other points, Akin argues that Cisco's specific beamforming requires three transmit antennas, yet the Aironet 1140, with a 2x3 MIMO configuration has only two.

We have a query into Cisco asking for comment. 

In our original story, we reported that, according to Cisco, 11n clients and 11n access points today already use something called explicit transmit beamforming, made possible in part because the client feeds information about the signal back to the access point, which uses it to optimize the signal. But 11a/g clients don't have this feedback capability. 

Yet they do get some slight performance improvement. That's because the access point's three RECEIVE antennas can use an algorithm called Maximal Radio Combining to adjust the different phases and amplitudes of the signals it receives from the 11g or 11a client on each of those antennas. There's an improved signal on the uplink, from client to access point, and therefore better throughput. 

According to Cisco, ClientLink uses the same MRC calculations to modify the signal on Aironet 1040's two transmit antennas, improving the signal "heard" by the 11a/g clients. 

In the Ruckus press release, the company's VP of Engineering, Steve Martin, argues that two transmit antennas limit the effectiveness of silicon-based beamforming.

"'Using only two Wi-Fi radios to transmit, vendors can perform only one of 802.11n's advanced techniques such as spatial multiplexing, transmit diversity or beamforming at any given time,' said Martin. "The limitation severely degrades potential system gains.'

"Martin noted that with only two antennas used to transmit Wi-Fi signals, as with Cisco's newly announced 802.11n access point, the theoretical signal gain with beamforming is limited to 3dB. This is the same theoretical gain achieved by the currently available transmit diversity technique. 'Ruckus doesn't have this limitation because we implement beamforming at the physical layer using smart antenna arrays on top of whatever 802.11 chipset is being used.'"

At a very non-technical level, this sounds like a sophisticated version of "We do it better." And Ruckus may indeed do it better. But that doesn't mean Cisco's beamforming does not or can not deliver a significant throughput boost for 11a and 11g clients.

One issue Ruckus does raise that's worth looking at more closely is whether, and under what conditions, Cisco's silicon-based approach could increase interference, and what the concrete results of that interference might be.

But assuming Cisco's ClientLink does indeed boost legacy Wi-Fi performance, I think it’s unlikely that most current enterprise wireless LANs and their users are going to see problems because it might not be the most elegant implementation. That, and much else, could change in cases where the enterprise WLAN is heavily loaded -- lots of clients using lots of bandwidth-hungry applications.

Join the Network World communities on Facebook and LinkedIn to comment on topics that are top of mind.
Related:

Copyright © 2009 IDG Communications, Inc.