Periodically varying channel gain mesurements

Hi,

I’m using Ubuntu 10.04 with a recent gnuradio-next branch and the UHD
driver on USRP2/WBX boards.
In an outdoor setup I have two receiving nodes about 40m away from
each other and from the transmitter. The transmitter is a PC with two
USRP2s connected with a MIMO cable. It is sending two different PN
sequences every 120ms on both interfaces at the same time. I correlate
the known sequences at the receivers to measure the channel gain. The
results can be seen here (x-axis is the packet number, y is the
absolute gain):

http://cs.ucsb.edu/~veljko/downloads/channel_gains_outdoor.jpg

The gain seems to be varying with around 2 Hz frequency. Since the
setup is static and the tx/rx gain is not modified during the course
of the experiment I’m almost certain that the artefact comes from the
system rather than from the environment. Does anyone have a good
explanation for this? Perhaps there is some sort of a gain control in
the USRP2 that I’m not aware of? I got similar behavior when I tried
the same experiment indoors.

Thanks,

Veljko

The gain seems to be varying with around 2 Hz frequency.

Quite possible, the WBX transmit has an analog power control loop
formed by the ADL5386. The “gain” setting on WBX transmit is really
more of a setpoint for the power control loop than gain

Jason

On Thu, Feb 24, 2011 at 16:54, Veljko P. [email protected]
wrote:

http://cs.ucsb.edu/~veljko/downloads/channel_gains_outdoor.jpg

The gain seems to be varying with around 2 Hz frequency. Since the
setup is static and the tx/rx gain is not modified during the course
of the experiment I’m almost certain that the artefact comes from the
system rather than from the environment. Does anyone have a good
explanation for this? Perhaps there is some sort of a gain control in
the USRP2 that I’m not aware of? I got similar behavior when I tried
the same experiment indoors.

This looks like the effect of timing skew between the transmitters and
receivers. Your correlation magnitude will vary in time with a period
related to how long it takes for the receiver sample times to “slide”
through one PN chip length. One way to test whether this is the case is
to
(temporarily) lock both the transmitter and receiver to the same
external 10
MHz reference; your correlation magnitudes should then be the same (with
some small noise variance.)

Johnathan

Thanks for the suggestions. I’ll see if the same 10MHz clock changes
anything. However, my initial idea was to have devices far apart.
Clocking the with the same ref clock in that setup seems impossible.

Cheers,

Veljko

On Thu, Feb 24, 2011 at 5:16 PM, Johnathan C.

This forum is not affiliated to the Ruby language, Ruby on Rails framework, nor any Ruby applications discussed here.

| Privacy Policy | Terms of Service | Remote Ruby Jobs