On 10/12/2020 11:32 PM, Xiang Ma wrote:
Think about this.
The signal transmitted at USRP TX is
image.png
Then after transmission, received at USRP RX is
image.png
Here, ignore the Doppler shift, and just consider the phase change due
to the transmission distance,
I know due to the wavelength, the phase received might be mod 2*pi,
however, how can we get the received below in USRP?
image.png
Should we use analog signals to get the phase offset? or After ADC?
Thx
Let us for a moment think about how radios actually work, rather than
how mathematical models of radios work.
You construct some signal at baseband--for simplicity a sinusoid. That
sinusoid is sampled at some rate, typically a rate that is a fraction
of the clock rate of the hardware. The basebanded signal is passed
through a special interpolating filter called a DUC, before being
presented to the DAC at the master clock rate. Once it leaves the
DAC it passes through an analog band-limiting filter, and then to
a mixer. The mixer has an LO at some phase that is random with
respect to the baseband signal. Once it leaves the mixer, it is amplified
and then presented to the antenna port, where it heads down a cable
or is propagated via an antenna--the new signal coming out of the
antenna or down a cable will have a phase that is a combination of
all of those stages in processing your baseband signal.
Now, the signal arrives at the RX, and the logical inverse of all of
that happens. Now, in the best of all possible worlds, the RX side is
a complete and perfect "mirror" of the TX side. The LOs are
perfectly synchronized (so you can eliminate them from the phase
estimates), as are the ADC and DAC clocks, and all the digital
transformations on the signal. Now, depending on the type of
USRP hardware you have, you *MAY* be able to synchronize the TX and
RX LOs. There are very few applications "out in the world"
that assume/require that the TX and RX frequency-conversion machinery
be perfectly in-phase and coherent. Radar is one of them.
But 99% of telecom type applications out there *necessarily* MUST
assume that the two "ends" of a conversation are not
mutually coherent with respect to all of the frequency-translation
machinery.
Now, there are comms applications that require that the TX machinery be
mutually-coherent with itself--because of MIMO, and this applies
to the RX as well. But what it DOES NOT require is that the TX and
RX be mutually coherent with each other, because the TX and RX
will be separate pieces of hardware, potentially separated by a
considerable distance, operated by different parties, and usually
different equipment.
So, in order to proceed further, it would be good to know, in a
high-level way, what it is you're ACTUALLY trying to achieve rather than
focussing on these small details. Because it's very easy to spend a
long time on these details and with a "bigger picture" it may be
easier and more expeditious to help you...
