Re: [Discuss-gnuradio] strange deterministic frame error in gr-ieee802-11

[Bastian Bloessl]

Hi,

On 30 Jan 2017, at 07:25, Nikita Airee <address@hidden> wrote:

I have a certain update regarding my issues and I'm hoping get some new perspective. 

I repeated my experiments randomizing the mac addresses and payload. You were quite right that the issue might be related to suboptimal waveform. Because sequence numbers which were always erroneous seemed to change. However, there were still certain numbers which were constantly wrong through the randomized frames. 

Then I found a Usrp 2922 which I believe is n210. Surprisingly, when used as a tx, 2922 doesn't give those deterministic errors and the total frame error rate too decreased drastically to around 1-1.5% from 4-5%. Also the constellation is visibly better.

great.

Is the value of fer acceptable at 1-1.5% or should there be absolutely 0 errors.

It depends totally on why you are setting up these experiments in the first place. If there is no interference from other devices it’s totally possible to reach 0%, but if it is required in your case, i don’t know.

Also could there be something wrong with my Rios?

If one SDR works and the other doesn’t it might be because the other SDR:

- needs different parameters

- uses a different/broken driver

- is broken

Maybe you could test the other SDR with different transceivers.

Best,

Bastian

Bests,

Nikita

On Jan 17, 2017 10:21 PM, "Bastian Bloessl" <address@hidden> wrote:

Hi,


On 01/17/2017 11:13 AM, Nikita Airee wrote:

    - Make sure that the xy plot shows clear constellations.

        tx_gain is set to 30dB and rx_gain to 0.


    If constellations don't look nice, you could try lowering the tx and
    increasing the rx gain. (also make sure that the antennas are
    connected to the correct ports).


The constellation is kind of clear(PFA the screenshot). Decreasing tx
gain and increasing rx gain doesn't help either.

I even tried repeating the experiment with a different set of x310s but
there was absolutely no change. Would you still say that it could be an
error in my hardware?

Also, another observation, the error in these frames seems to be around
the same area of that particular frame everytime. Could this mean something?

Unfortunately, I don't have the hardware to try.

Since you always loose the same frames it sounds like a bug in the receiver, but then they should also be dropped in  simulations.

The last idea would be that this particular frames are mapped to a suboptimal wave form (high PAPR?) The constellations already look like the gain is too high. Maybe frames with high peak amplitudes get dropped.

Best,
Bastian

--

Dipl.-Inform. Bastian Bloessl
Distributed Embedded Systems Group

University of Paderborn, Germany

http://www.ccs-labs.org/~bloessl/