Channel bonding revs WLANs past 100 Mb

Using a number of techniques, Atheros' Super G and Super AG technology has enabled a doubling of data rates beyond the industry-standard wireless-LAN maximum. The technology has four components that allow this superior performance, all of which are interoperable with standard 802.11 products. They are demand-driven and engage only when there is a real need for throughput; otherwise the product is standard 802.11.

First is a Lempel-Ziv compression engine embedded directly in the silicon so that it can increase throughput at line speed without the usual performance drops associated with compression algorithms running on the host processor. Bursting and fast-frame developments have also been devised in order to enhance the signaling protocols to use shared radio spectrum more efficiently. These three developments alone realize a 50 to 70 percent increase in throughput.

A fourth technology, called Turbo, is used to raise data rates above 100 Mbits/second when required by the traffic demand. This is a channel-bonding mechanism that uses extra bandwidth for the signal to raise the data rate. This capability is automatically engaged as required (as is a turbo in an automobile) and can be accessed under user control in the software. It also looks for other users in the channel, to ensure that Turbo is not engaged when an incompatible product is in use. In that case the product will continue in standard 802.11 operation.

One of the major advantages of Super G comes to light in the context of temporal diversity. Wireless always uses a shared medium and as such, Super G actually helps to limit overall network interference by reducing the airtime that stations require to transmit data, thereby increasing performance for others in the network. In addition, a wall can reduce the strength of a signal by 10 to 20 dB or more. This dramatically enhances the effect of any temporal or frequency diversity that exists and reduces any overlap to a level of insignificance.

But a side effect of the success of Super G is that some in the industry have stopped considering how people use such technology in the real world, and instead manipulate non-real-world lab data that fails to map to real-world conditions

Low-power WLAN products designed for the unlicensed band have a considered use built into them, but while lab testing has some comparative performance value, it is not a model of the real world. People do not live in labs and do not operate these products in the way some recent tests have proposed. Real-world scenarios include houses and apartments with walls and space separating them. This is an element that cannot be disassociated from consideration and was part of the initial concept of using unlicensed frequencies.

The unrealistic environment of testing uncoordinated networks in close confines produces problems that are universal. It is easy to show two standard 802.11g products interfering with each other at opposite ends of a lab bench when streaming video. This is the side effect of using them in an unrealistic scenario. Ironically, in the real world outside the lab, the opposite problem is true: One rarely sees complaints about an 802.11 signal's being too strong or interfering; the most common complaint is that range is not good enough.

Atheros is a strong supporter of the IEEE standards process and Wi-Fi Alliance interoperability initiatives. We did considerable testing of ours and other products both before and after we developed this technology. We continue to see standard products interfering in unrealistic-use cases, such as lab testing, and we counsel everyone to use common sense when conducting such tests. The intended use is as important as the product.

Sheung Li is product line manager at Atheros Communications (Sunnyvale, Calif.).