Re: [Discuss-gnuradio] signal cancellation example using GRC and USRP B210

[Inkyu Bang]

Dear Marcus,

Some questions on signal cancellation example again,
(I have tried other things and am back to this topic again.)

Actually, what I try to do is a signal cancellation in the air.
(For example, interference cancellation in the air,
so a specific receiver will experience almost zero interference).

Could you make some comments on the following experiment what I did?

One transmitter (USRP B210) and one receiver (USRP B210)
Carrier freq.: 5.89G
signal: sine wave having 1000 Hz frequency
sampling rate: 5 Mbps
office environment / LOS (1m) / assume channel is quite static

1. Sending a signal from TX to RX and recording it into a file (binary data).
Since I turned on RX first, the file has some zero values (not really zero, noise level signal) at the beginning of the file.

2. I repeated the experiment one more and save the samples into another file.

3. Using some tools (GRC / MATLAB), I read both signals (say, A and A') and make time synchronization between them by removing zeros at the beginning of each signal.

finally, "result" = A - A'

I expected to see almost zero signal (noise level) but still, signals are not canceled.

Could the frequency offset be time-variant whenever I tried experiments even if I used the same hardware? (freq_offset1 at t1 / freq_offset2 at t2)

Do you think any other reason for that?

Thank you,
Regard,

Inkyu

2018-04-29 18:06 GMT+08:00 Müller, Marcus (CEL) <address@hidden>:

Hi!
On Sun, 2018-04-29 at 01:17 +0800, Inkyu Bang wrote:
> Dear Marcus
>
> Thank you for the comments.
>
> I am trying to make a beam which is canceled out (i.e., null) at a
> specific location but is not elsewhere.
>
> I am aware of beamforming and half-wavelength things
> (By the way, a wavelength of 6GHz is 5cm)

oooops :D

> So I put two Tx antennas and one Rx antenna in proper locations to
> satisfy the condition which you said (i.e., half wavelength distance
> difference).
>
> I do not know much about phase offset and USRP hardware things.

So, your two TX chains have an unknown phase offset. You could actually
find that by, for example, sending the very same signal on both
antennas, and measuring the angle of the nulls :)

>
> Can I ask a few more questions?
> (It might be silly questions.)

No such thing as silly questions!

>
> I used USRP B210 which has two RF chains and it seems to use the same
> clock for those two RF chains.
>
> Here is a link to brief specification of B210:
> https://www.ettus.com/content/files/b200-b210_spec_sheet.pdf
>
> Q1) In this case, is there a chance of different phase offsets
> between two antennas from the same USRP?

Yes, there is. The clock's the same, but not necessarily its phase.

>
> Q2) Then, how can we calibrate those different phase offsets between
> two antennas in the same USRP?
> (What happens in MIMO transmission in real SDR devices?)

For example, through geometric observation as described above, by some
kind of cable loopback (but that would require phase-length calibrated
cables and some way to switch over to the antennas afterwards), or by
techniques that would send two different (optimally: orthogonal)
signals, one from each TX antenna, and using math after the one RX
antenna to tell the phase difference of these two. A classic for that
would be to use Gold codes, but for your use case, basically any good
pseudorandom bit sequence would do (if you can construct a sufficiently
orthogonal counterpart, but that might be as simple as XOR with
10101010...).

Best regards,
Marcus

>
> Thank you.
> Regards,
>
> Inkyu
>
>
>
> > 2018. 4. 27. 오후 10:13, Müller, Marcus (CEL) <address@hidden> 작성:
> >
> > Dear Inkyu,
> >
> > there's no guarantee that the phases of both TX are exactly
> > identical.
> > In fact, you should expect an unknown offset.
> >
> > Also, you'll notice that your two antennas are at two different
> > positions in the room. You're accidentally building a beamforming
> > system! So, these nulls will not be everywhere in the room, but
> > only in
> > a specific direction in the far field of the TX antenna. And with a
> > carrier frequency of nearly 6 GHz, a wavelength in air is
> >
> > c/f ≈ (3·10⁸ m/s) / (6·10⁹ 1/s) = 0.05 m = 50 cm
> >
> > and half that distance from a (single) transmit antenna would be
> > the
> > distance at which two received signals would be exactly of opposite
> > phase.
> >
> > So, I'm not sure what you want to demonstrate, but unless it's
> > beamforming, you're probably not doing it right :) Which is no
> > shame,
> > but it would be interesting to hear what it is that you want to
> > achieve.
> >
> > Best regards,
> >
> > Marcus
> >
> > On Fri, 2018-04-27 at 20:44 +0800, Inkyu Bang wrote:
> > > Hi, all !!
> > >
> > > I am trying to make a simple example of signal cancellation using
> > > GRC.
> > > However, I did not get any good results.
> > >
> > > Here is my setting
> > >
> > > Plan: sending two sine wave with different phases and receiving
> > > canceled signal (only noise)
> > > Center frequency: 5.89GHz (I am using the antenna supporting this
> > > center frequency)
> > > Sampling rate: 5MHz
> > > Frequency of sine wave: 1kHz
> > >
> > > USRP: B210
> > > Tx GRC flow graph: signal source block + USRP sink (two channels)
> > > Rx GRC flow graph: USRP sink + QT GUI time sink
> > >
> > > When I send two sine signals, I used two RF chains in the same
> > > USRP.
> > > I received those signals at the other USRP.
> > >
> > > I expected to see canceled signals (i.e., only noise level power
> > > in time domain)
> > > but still, I see sine wave signal.
> > >
> > > I thought frequency offset in the receiving USRP is applied to
> > > both sine waves.
> > >
> > > Do I need to consider frequency offset first?
> > > Could anyone help me to solve this problem?
> > >
> > > Thank you.
> > > Regards,
> > >
> > > Inkyu
> > > _______________________________________________
> > > Discuss-gnuradio mailing list
> > > address@hidden
> > > https://lists.gnu.org/mailman/listinfo/discuss-gnuradio
>
>