Concentrating a wireless signal in a tight beam has tremendous advantages. You need less power for equivalent distance, you create less interference to others and, if your receiving antenna is similarly directional, you avoid a lot of interference from others. Of course people have been using directional antennas to achieve narrow wireless beams for years. But a directional antenna has to be aimed and then it's fixed.
That's great for a point-to-point link but useless for someone on the move.
Beamforming creates a narrow wireless beam that can be steered in software. Suddenly the advantages of narrow beams can be applied to point-to-multi-point systems and in systems where one or both ends are in motion. Of course the process is quite sophisticated and, until recently, was very expensive. Luckily that is changing, rapidly!
This slide shows how, with beamforming, one Wi-Fi access point can communicate, in rapid succession, with three different clients. And this can happen despite another access point using the same channel to talk with one of it's clients. Indeed, it all works despite the existence of a legacy Wi-Fi device that is broadcasting omni-directionally on the same channel. Pretty powerful!
Proprietary or "standard" ??
Based on questions I get, there appears to be quite a bit of confusion. The answer of course is we see both proprietary and standard implementations within the Wi-Fi market.
Beamforming is an optional part of the 802.11n standard and silicon support for this option is emerging (more on that in a moment). Beamforming is also part of the most recent revisions of WiMAX and LTE standards. Confusion arises because there are several different things which are legitimately called beamforming.
The simplest is a switched beamformer in which there are multiple directional antenna elements and the radio or radios are connected to the appropriate elements as needed. This is what Ruckus Wireless does today. It's what Vivato Networks did with 802.11g back in 2003. The Ruckus products have 12 or more fixed elements and, on a frame-by-frame basis (e.g. every 10 ms), they decide which two of those antenna elements to connect to the two terminals on an Atheros (2x2 MIMO) Wi-Fi chip, based which combination works best for the specific device the access point is communicating with. Many people have patents here, but the ideas are very old, so the patents may not be very valuable.
Next are phased array beamformers. Here multiple simple antenna elements, equally spaced, form an array. Each element sends (or receives) the same signal but phase delays are introduced so, via constructive and destructive interference, the resulting wireless signal forms a narrow beam. This beam can then be steered by varying the phase delays. Phased array beamformers are also well established with examples dating back to 1905 and widespread use in radars since World War II.
Finally, in MIMO systems, precoding is a mathematical generalization of beamforming that leverages antennas elements that may not be equi-spaced and deals with the fact that the other end has multiple receiving antennas. And at the receiver end, maximal-ratio-combining (MRC) optimizes decoding of the transmitted signals. Very roughly you can think of these MIMO computations as putting the maximum beam lobe as close as possible to the desired target while placing nulls (of the antenna pattern) as close as possible to the primary interferers.
Today's widely deployed Wi-Fi chips have two radio chains (2x2 MIMO) which are typically used for horizontal and vertical polarization, but 3x3 and 4x4 MIMO chips have come to market. With 4x4 MIMO Wi-Fi chips we see transmit beamforming via precoding and receive beamforming via MRC. Indeed, two silicon startups, Quantenna Communications in California and Celeno Wireless in Israel, have announced Wi-Fi chips that support 4x4 MIMO with transmit beamforming and both vendors' chips have shown up in consumer products, e.g., the Quantenna chip in the Netgear WNHDB3004.
The 802.11n standard specifies how the needed information is passed, so the computations that Quantenna and Celeno do (and others in the future) can be carried out at either end even though the devices are from different vendors.
Beyond mentions in my speaking engagements and wireless tutorials, beamforming will be directly useful to netBlazr, our new (and radically different) wireless ISP. netBlazr depends upon participants deploying network elements. Today, these require aiming but, with beamforming, aiming won't matter and, if the unit is bumped later on, beamforming will compensate on-the-fly.