kev <invalid@invalid.invalid> hath wroth:
>Jeff Liebermann wrote:
>> kev <invalid@invalid.invalid> hath wroth:
>>
>>> "Beamforming is done by multiplying the signal at each antenna by a set of
>>> complex (phase and amplitude) weights. This weighing is implemented per bin
>>> in the frequency domain. In transmit, the beamforming weights are set so
>>> that the signals from the different antennas sum up coherently at the
>>> receiving client. In receive, the weights are set so that the signals from
>>> the different antennas sum up coherently at the beamformer output."
>>> http://www.wavion.net/tech/beamforming.html
>>
>> Yeah, I saw that. My question is how they managed to get BOTH 8dB of
>> processing gain, and 8dB of direction beam forming gain out that that
>> array. Very roughly, you get a 3dB gain boost every time you double
>> the size of an antenna array. So, you start with one reference
>> antenna at 7.4dBi gain.
>> 2 antennas = 10.4dBi
>> 4 antennas = 13.4dBi
>> 6 antennas = 14.8dBi
>> That's an increase of 7.4dB gain over a single antenna, which is a bit
>> less than the claimed 8dB gain. There's no compensation for losses
>> and phasing errors. I don't think 8dB beam forming gain is possible.
>They and the FCC are using a different method.
>Array gain is the gain provided by coherently combining the beamformer's
>N antenna signals.
Ummm... that's exactly what I did. I wanted to derive the method, so
I combined the output of all the antennas. Comparing results:
10*log(6) = 7.8dB gain.
7.4dBi (single antenna gain) + 7.8dB = 15.2dBi array gain.
Close enough.
>In receive, this gain is given by 10·logN when the signals at the
>different antennas are equal, and slightly lower when the signals are
>not equal due to the effect of multipath. However, 10·logN is a good
>approximation for all practical purposes. This implies that for a six
>element antenna array (N=6), the array gain is about 8dB.
What I'm ranting about is that I don't think it can be done in
practice. Nailing the phase of 6 antennas, through 6 allegedly
identical transmitters, and conglomerating the resultant decoded data
is not easy. It's not just matching the assorted digital and analog
delays, it's also the susceptibility to multipath and moving doppler
sources. I would be really impressed if it could be made to work.
>In transmit, the array gain depends on the total power transmitted. If
>the power transmitted from the array is equal to the power transmitted
>from the reference single-antenna, then the transmit array gain is equal
>to the receive array gain, i.e., 10·logN. If, however, each antenna
>element in the array transmits the same power as the reference
>single-antenna, then the total power transmitted from the array is N
>times that of the reference antenna. In this case the array gain is
>given by 20·log N. So for a six element array, the array gain is 16dB.
In transmit, in order for this to work, the system has to assume that
it knows at which azimuth the receiver is located. If it's off angle
somewhat, the combined arriving signals get smeared with phase
distortion and possibly (not sure) inter symbol trash. In other
words, I wanna see it work under realistic conditions.
>http://www.wavion.net/tech/linkGain.html
Thanks. I see how it works. I'm just suspicious if the numbers are
achievable.
--
Jeff Liebermann
jeffl@cruzio.com
150 Felker St #D
http://www.LearnByDestroying.com
Santa Cruz CA 95060
http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558