Last Post: December 9, 2010:
How much bandwidth can you get out of a 10Mhz (or 5Mhz) 4.9Ghz Channel using 3x3 MIMO? We are using Firetide mesh access points. We have heard the true throughput of 22Mbps, but haven't seen it posted anywhere or reference to where the value came from.
The only thing we have seen from them is 180Mbps on a MIMO 40Mhz channel, same as the 40Mhz 5Ghz channel-bonded setup. I want to know what the 10Mhz and 5Mhz channels are capable of.
There are only 5 10Mhz non-overlapping channels in the 4.9Ghz spectrum, 2 20Mhz and 1 40Mhz channel. You have to pick between 5, 10, 20, or 40Mhz with MIMO.
Although a lot of math could get involved, anytime you cut the bandwidth in half you lose half your throughput. Technically the larger the channel the more efficient it becomes, but it is only slightly.
There is another aspect to narrower channel widths to consider. The narrower the channel, the better range you will get.
Ruckus Wireless - The leading seller of outdoor mesh APs in the world. :)
Is the greater range due to the same amount of power, not being so spread out ?
Are the allowable power levels different ?
Thanks for the assistance.
Here's what I have figured out (or "think I've figured out).
The 4.9Ghz public safety is almost the same setup as standard 5.0Ghz and uses the same or closely related chipsets and radios. (In Japan the 4.9Ghz range is part of the 5.0Ghz spectrum.) The differences are channel restrictions are a bit tighter and you can split the 20Mhz channel in half or in a quarter. I believe power levels are the same as well.
Per 802.11-2007: 10MHz and 5 MHz is also called OFDM 'half-clocking' and 'quarter-clocking' (as opposed to bonded channel for 40MHz). This means that 5, 10, and 20Mhz have the same number of channels.
20MHz 52 subcarriers (48 data, 4 pilot)
40MHz 108 subcarriers (102 data, 6 pilot) or slightly more than double (hence the slightly more than double bandwidth)
http://standards.ieee.org/getieee802/download/802.11-2007.pdf page 673
I can't find much on is how MIMO works on 5 and 10MHz and how it splits it into steams (20 and 40MHz split into 4 streams.)
BTW: the advantage of the 5 and 10Mhz channels is that you have more channels in a limited space, sacrificing bandwidth of each channel. We're trying to figure out how much is lost. (My _guess_ is 20Mhz to 10 will drop slightly more than half.)
Again, thanks for your time.
WLANMan: It isn't a TX power thing, it is a noise reduction advantage. Imagine you are a dog. You can now hear frequencies that you could never hear before. The problem? You now hear natural noise (in RF it is called White Gaussian noise) that you could never hear before, thereby reducing your SNR.
Conversely, if you narrow the available bandwidth to a very narrow set of frequencies, you hear less noise and increase your SNR. In my few WISPs over the years narrowing the channel width has been a great trick to getting to customers that we could never reach before.
jbdkaty: I've emailed one of our founders that basically helped invent MIMO for Wi-Fi and he confirmed what I'm about to say as I wasn't 100% sure.
MIMO will work fine on a narrower channel width. Yes, 11n does specify up to 4 spatial streams but 2 is the most common and 3 streams are coming out in the mainstream but not really on the market yet. Keep in mind that the first two numbers (3x3) doesn't actually tell you how many spatial streams there are. There is a third number that tells you the number of spatial streams. For example, our current outdoor 11n AP is a 3x3:2 which means 3 transmit, 3 receive and 2 spatial streams.
When cutting your channel width in half you lose a bit over half the throughput but it is so minor that it shouldn't play into your decision.
I do however encourage you to look further into using the 4.9 band. It really isn't all it is cracked up to be. It doesn't provide for much total throughput and it is no more secure than 5 GHz. It sounds exclusive but in a decent size metro when every emergency service can use it, it's exclusivity to emergency services diminishes quite a lot.
We have found that many emergency services transport their data in 5 GHz because it provides more available throughput and a wider range of equipment at better prices.
Seems obvious now that you've explained it.
The main parameters which affect the ?half-bandwidth? improvement are:
1. Appropriate filtering at the receiver ( prior to the demodulator )
2. Normally maintaining the same Tx power as the ?full bandwidth case? ( although this does not always have to be the case ).
When radio signals enter into a receiver, a filter removes signals and noise below the lower passband frequency and above the upper passband frequency. We end up with a modulated signal ?sitting on top of?a ?noise floor?.
Noise can come from a variety of sources?atmospheric noise, electronic noise from the components that make up the radio receiver etc.
We refer to the greatest contribution to the noise as thermal noise. Any object above absolute zero will radiate noise, as electrons are raised to higher energy levels and then ?suddenly drop down? to a lower energy level giving off electromagnetic radiation. This occurs in a random manner. When we can do this in an organized manner with the appropriate materials we can get laser radiation.
Noise power ( although tiny ) can be measured in units of Watts ( tiny, tiny fractions of Watts ) or in dBm or dBW. There is an equation for noise power given by:
N = kTB
Where N = noise power in Watts
K = a constant called Boltzman?s Constant ( 1.38 x 10 minus 23 Joules per degree Kelvin )
T = Noise temperature in degrees kelvin ( Not quite the same thing as physical temperature, but related to it )
B= Noise Bandwidth in Hz ( this is what we are concerned with here )
If we physically alter the surrounding temperature of a radio receiver, we can alter the amount of noise produced ( less ?activity? from the electrons ). NASA does this with the Deep Space tracking stations by using liquid helium to physically cool the low noise amplifiers.
One of my first jobs was working on cryogenically cooled LNAs. They had a refrigeration system pumping the helium and carbon filters to remove impurities. If you looked on a spectrum analyzer ( which by the way measures S+N/N and not S/N directly...microprocessors usually then apply a correction factor to give true S/N in some cases ) you could see the noise floor plummet when the chillers were activated. They were quite temperamental and if an impurity got into the cooller loop you could actually see a "bump increase" in noise every time the impurity appeared at the right spot in the loop.
For our purposes we can assume that the noise temperature remains fairly constant ( very small changes from a hot office to a cold warehouse ).
Boltzman?s constant is exactly what it says it is?a constant.
So now we have effectively two constants in our equation?.k and T.
So, as we reduce the noise bandwidth B, we can reduce the total amount of noise entering the demodulator. How do we do this ? By selecting a lower bandwidth on the receiver and transmitter. The actual signal is now reduced in bandwidth by say half.
The famous S/N or signal to noise ratio equation can now be written as:
If we keep the same signal ( TX power, assuming no changes in path conditions ) as when we had ?full bandwith?, we can see that if the value in the denominator ( kTB ) has been reduced, the overall value of S/kTB increases.
This is just one of several methods of ?noise reduction?.