Regarding calculating the noise power

Hi.
I am using USRP2 testbed. I have been recently working on plotting the
PSD of the received signal and I was able to do it. The graph plots
the signal in baseband frequency level.
Then I turned on the receiver node while keeping the sender node off.
When I plotted the PSD graph for that, I was expecting to see a PSD
which should be more or less constant over all the frequencies.
But I saw a peak in the 0 Hz which is the center frequency. I was told
that this is due to DC offset. I am not sure what is meant by that
exactly and how to control or measure the DC offset in the output I
got.
Can anyone help me out?

Also I have been working in a modified program of benchmark_rx for
this case. It would be more helpful if someone helps me to reduce it
or to measure it, so that I will be able to explain to the people who
will ask me about the spike.

On Wed, Jul 6, 2011 at 4:44 PM, shantharam balasubramanian <
[email protected]> wrote:

Also I have been working in a modified program of benchmark_rx for
this case. It would be more helpful if someone helps me to reduce it
or to measure it, so that I will be able to explain to the people who
will ask me about the spike.

Yes, you will see a DC offset in these receivers. You can think of
removing
it as subtracting off the time averaged value of the signal. I think a
better solution, though, is the block gr_dc_blocker_{cc,ff} that I
recently
added to the project. It’s a narrow notch filter at DC.

Tom