The type of interference in question here is called multipath. It's pretty much what the name says. A radio signal from a transmitting antenna takes multiple paths to the receiving antenna. The signal we normally want is the one that flies like the proverbial crow in a straight line from transmitter to receiver. Those other paths come from the same transmitted signal bouncing off hills and buildings outdoors and walls and objects indoors.
You know that anytime waves of any type meet, they interfere creating hotspots and dead spots. A common experience of this is when you pull up to a stop sign and the FM station on your car radio gets distorted or cuts out. If you can creep a foot or so forward, it comes back in. That's multipath in action. Since the wavelength of an FM radio station is only about 3 feet long, moving a foot makes a big difference between the multipath signals adding and subtracting.
Wireless networks are built on radio transmitters and receivers. The same multipath effects occur at 2.4 GHz as they do at 100 MHz. The only thing different is that the wavelength of a 2.4 GHz signal is about 5 inches instead of 3 feet. Indoors there are lots of things for such a high frequency signal to bounce off. It's likely that your wireless router and notebook computer are bathed in a sea of multipath signals of varying strengths.
A classic way to reduce multipath problems is called spatial diversity reception. Two receiving antennas are used, separated by inches or feet. The receiver compares the signal strengths coming from each antenna and picks the one with the strongest signal. That helps, but you can still have strong and weak spots in any area. What if you could take advantage of that to get more bandwidth?
It sounds a bit counterintuitive to expect that interference can help increase bandwidth instead of reduce it. But that's what "smart antennas" using a technique called MIMO or Multiple-Input Multiple-Output do. The boost comes from transmitting two different streams of data on the same channel using two transmitting antennas. The resulting cacophony is sorted out by two receiving antennas and some clever processing algorithms. For instance, if a WiFi access point normally has a bandwidth of 54 Mbps, it could have as much as 108 Mbps for two independent streams of data.
What we normally get when there are two stations on the same channel is a jumbled mess of both signals or the stronger of the two overpowering the weaker one. But when you factor in the multipath cancellations that you can't avoid anyway, the two receiving antennas will not pick up both datastreams equally on both antennas. One will be stronger at the first antenna. The other will be stronger at the second.
The overall technique is referred to a spatial multiplexing rather than simple spatial diversity reception. The multiplexing refers to the inverse multiplexing that comes from combining two datastreams into one larger one to effectively increase bandwidth.
Does it work? You bet it does. It works so well that the IEEE is likely to include MIMO when it approves 802.11n, a WiFi standards upgrade that increases wireless data rates to 100 Mbps and beyond. As usual, manufacturers are getting out ahead of the standard and releasing pre-802.11n hardware. First was Belkin Pre-N Wireless Networking. Other companies have also released "pre-n" equipment with the benefit of an immediate speed and coverage improvement, backwards compatibility with existing "b" and "g" WiFi, and at least a hope that only minor upgrades will be needed when the official standard is released in a year or two.
