Re: [Discuss-gnuradio] Possible bug with PATH/gr-channels/lib/cfo_model_impl.h

[Tom Rondeau]

On Wed, Feb 5, 2014 at 7:40 PM, Aditya Dhananjay <address@hidden> wrote:
> Dear All,
>
> I suspect there is a bug in cfo_model_impl.h, where calling set_std_dev to
> set the std_dev back to 0 causes undesirable effects.
>
> How to reproduce the bug:
>
> a) Standard OFDM TX-RX example in GRC. Connect the TX block to the RX block
> through this channel model block.
>
> b) Use a variable slider to control the std_dev of the CFO model block.
>
> c) Start the flowgraph, with 0 as the initial value of the std_dev. All
> should be fine.
>
> d) Bump up the std_dev and observe high packet errors.
>
> e) Drop the std_dev back to 0. While we expect the packet error rate to drop
> back to 0, but that does not happen.
>
> Reason: The value of d_cfo might be a high non-zero value. Even though we
> drop std_dev back to 0, d_cfo remains high, and continues to create CFO.
>
> How to fix it:
>
> In the function set_std_dev, reset the value of d_cfo to 0.
>
> If someone could verify that this is indeed a bug, that would be very
> helpful. Thank you in advance for your help.
>
> Best regards,
> Aditya


Nope, not a bug. You're reducing the standard deviation, but the
actual CFO value is not this value but is adjusted based on it.

I think more of what you are looking for is a direct way to set the
value of the CFO (or something like a reset() function).

Tom