I'm not certain, but I think the word channel may be misleading here. I think that MIMO is actually using the same bandwidth, just combining multiple RF paths to enhance the signal to noise ratio. Another MIMO link is here. [networkworld.com]

That a great way to be a good neighbor. Piss all over all the channels available so no one else can do anything.

I was about to post with the same sentiment until I read more on the tech.

MIMO is not your typical blast-it-on-multiple channels approach. This article [thechannelinsider.com] discusses the technology. Instead of using up a bunch of channels, MIMO systems send multiple signals on one channel and use multiple antennas and advanced algorithms on both ends to sort out which signal came from or went which direction.

4x the speed is still not that great.

Current 802.11g devices have a theoretical throughput of 54mbit, and a real-world throughput of actual data of 10 to 20mbit. So it follows that Airgo's new cards will permit 40 to 80 megabits.

Now, wired 100mbit networks can reach 80mbit real-world speeds (Actual after-overhead bandwidth), so at first glance it looks like we're there. Except we're not.

The important things to keep in mind is that wireless networks behave like hubs, not switches, and on top of that all data must go through the access point. So if you have two computers close to an AP, you take up 40mbit for computer -> AP, and the other 40mbit for AP -> computer.

In other words, they claim 240mbit, but the fastest real-world transfer between two wireless devices is probably about 40mbit, IF those computers are very close to the access point. If the computers are a bit further away, you will get 20mbit. 4 computers doing 2 transfers and each transfer goes at 10mbit.

So you see? 4 computers 50 feet away and you're already down from 240mbit to 10mbit. This is very far away from wired performance.

To put this in software terms, They are multithreading the connection.

-Rick

54Mbit/sec = 27 Mbit/sec each way. 20mbit/sec is the max reasonable bandwidth...

Yep... and I've had one for a couple of months. Although I'm using it with 802.11b and 802.11g devices, I've noticed a big performance boost. The major advantages:

• The ability to go really fast
• No mixed mode (i.e. an 802.11b device on the network doesn't slow down 802.11g
• Greater range with existing 802.11b and 802.11g devices

I think you're right. The top level post has everyone all confused because they used the phrase "multiple channel" instead of multiple paths. I replied to the original post to that effect, but a lot of people are still missing it.

Since the net effect of the technology is to create a higher gain antenna system, via electronic phased array, it is actually a less interfering signal than normal.

Airgo is a participant in one of two consortiums of companies promoting competing technologies to use in the 802.11n standard. Here's an article that covers the situation:

http://www.reed-electronics.com/electronicnews/art icle/CA445702 [reed-electronics.com]

Airgo is obviously trying to gain leverage with their technology by getting it out on the market early. I don't think this is a good thing in the long run, since we all have benefitted by the degree of standardization in 802.11b/g and Airgo seems to be trying to get their own proprietary technology out there in front of the legitimate standards process.

-R

There's an article on MIMO [technologyreview.com] in the latest physical issue of Technology Review [technologyreview.com] magazine. Fortunately, the article's on-line.

I design wireless networks and hardware for a living. I'm running simulations of video streams across ns2 and NCTUns as I write this.

Yes, you can increase throughput. Yes, you can cross-correlate FEC across channels to reduce errors. However, this solution hogs the spectrum, isn't tileable to create large wireless networks because of its inefficient use of channels (not to mention that the algorithm they're probably using only works because 802.11 has fairness problems, will definitely conflict with 802.11n (which also uses MIMO), and has a kook for a CEO.

"When MIMO was first unveiled, it reversed over 100 years of scientific thinking by harnessing natural radio wave distortions, which were previously perceived as interference, to deliver dramatically increased speed, range, and reliability," said Greg Raleigh, chief executive of Airgo. "With True MIMO Gen3 technology, our team has achieved a scientific milestone by proving that wireless can surpass wired speeds."

This guy is talking about something no more complex than using four radios at once and he's talking like it's the Second Coming. Could someone please bonk him with a hardbound copy of the 802.11n standard?

I like the idea of tying into multiple access points to increase throughput, but because their method relies on inherent 802.11 unfairness in order to work, I can't see this working in a large deployment.

This is pre-802.11n stuff, folks. Wait for the real stuff to come out from established vendors who actually contributed to the standard, instead of these guys who seem to be trying to break everything else by layering their solution on top of 802.11a/b/g, disrupting it in the process.

You're thinking of just taking different frequency channels and bonding them together. That basically just uses more frequency to transmit more data rate, while the number of bits per second per hertz (the spectral efficiency) remains the same.

In the MIMO system they're discussing here, you use the same frequency bandwidth but deploy multiple antennas, which gives you spatial diversity. Wireless communications are basically limited by the probability that your channel goes screwy and experiences what's called a fading event, where your signal suddenly drops because of interference. This means you have to be more conservative in the data rate you transmit at.

What they're trying to do is transmit, receive, and resolve multiple signals in the same frequency band by using multiple antennas, and resolving them in a clever way to try to create independent data channels. Since each antenna is physically at a separate location, the signal paths (and hence the fading characteristics) from the transmitter to the receiver will be more independent. Then the odds that all channels experience fading simultaneously drops significantly, improving the overall robustness of your communication channel to fading. That means you can be less conservative and achieve higher bit rates through your channel.

In short, same frequency usage, but they're getting spatial diversity by using more antennas and giving themselves a more robust channel.

Retailers are always putting accesspoints and wireless NICs on sale for less than $5 each (like CompUSA [compusa.com]). That makes your cost $10 to go wireless; I don't actually consider that an "investment".

You can switch between wired and wireless operation depending on your bandwidth/mobility needs of the moment, so you're not giving anything up.

"Up to" means that there was a second in which they may have reached 240 megabits per second. I'm not sure if this is the data before or after compression, so 240 bits of 1's run-length encoded might easily be transmitted in one second.

It is also unclear as to whether the data was actually intact or not, how much error-correction the network card needed to perform, how many resends were required, etc.

In other words, even a transmitted rate of 240 megabits per second need not equal 240 megabit transfer rates. There are plenty of ways to fudge the numbers.

A trivial example: A network card operates at 240 megabits per second, but needs 240 retries to get enough data across for a genetic algorithm to build the most probable originating packet that could produce the data received, where the genetic algorithm adds several minutes to the transmission time of a single packet. At what speed does the card operate?

That would be under Title 47, Chapter 1, Part 15, Section 15.5, Subsection B of the US Federal Code.

http://edocket.access.gpo.gov/cfr_2004/octqtr/47cf r15.5.htm [gpo.gov]

This isn't a channel bonding scheme. The article is a bit misleading when it says that MIMO is"a wireless technique that uses different radio channels to improve both speed and transmission quality."

Yes, it's using different channels, but not in the sense you and I typically think of it, as chunks of spectrum. Rather, it is exploiting multipath, where each path is treated as a separate channel. Multipath is usually a significant problem for traditional wireless communication, because it causes dropouts and frequency nulls and such. But it turns out that if you're clever (and these guys are) you can exploit multipath to shove more data down the same sized pipe.

So in short, they're not hogging spectrum to get these speeds; they are being good neighbors.

Actually they're parallel processing the channel using arrays of virtual directional antennas, rather than multithreading (which has some degree of time-slicing implied, even with multiple instruction processors sharing data operation execution units).