Hi Kamran,
It's been a while since I last updated my code. The last time I ran my tool
it generated around 52K/s samples (so a little bit slower then I previously
claimed).
This was with a WNDA3200. It uses an AR9280 wireless chip, where setting
the highest bit will make it continuously generate samples ("never
disable"). Firmware was modified to pass spectral samples as packets. The
driver passes the samples to userland, again as simple packets. I then made
a tool in C++ which captured these packets (so the spectral samples) and
wrote them to file. Running it for one second returned 52K samples. Code is
based on compat-drivers-3.9-rc4-2su and used with Linux kernel 3.7.2. I was
hoping to use this for my own research one day, but I haven't continued
with it yet.
I never fully understood the difference between the `ss_fft_period` and
`ss_period` parameters though [1]. Maybe someone can clarify that. Also,
setting `ss_fft_period` to either 1 or 0 did not speed up the sampling
process. Samples were always 17-25us apart (according to the MAC timestamp
of the wireless chip). At the time I wasn't able to generate samples in
quicker succession. Perhaps a limitation of the wireless chip?
Scanning at the symbol level seems difficult. If more coarse grained
measurements are also acceptable (say 4us samples every ~20us) we can
perhaps collaborate on the code? I should be able to get it all working
again, and update it to work with the latest kernel! :)
[1] https://wiki.freebsd.org/dev/ath_hal%284%29/SpectralScan
On Mon, Apr 7, 2014 at 8:35 PM, Kamran Nishat <kamran.nis...@gmail.com>wrote:
> Hi Mathy
> Can u please share the details of your 60K/s samples with us. How were u
> able to do that (some setup and variable settings). Did u get them in
> debugfs entry or any other way.
> What we want to do is to get OFDM samples at every 4us to scan at the
> symbol level.
> Kamran
>
>
> On Sun, Apr 6, 2014 at 1:47 AM, Mathy <vanho...@gmail.com> wrote:
>
>> Hi Kamran,
>>
>> Some time ago I was able to get 60k samples per second using a modified
>> ath9k_htc driver and firmware. I don't know whether better is possible.
>> What are you planning to do with it?
>>
>> Cheers,
>> Mathy
>>
>>
>> On Sat, Apr 5, 2014 at 10:15 PM, Kamran Nishat
>> <kamran.nis...@gmail.com>wrote:
>>
>>> Hi,
>>> What will be the finest granularity of time to get spectral scan. I
>>> assume it should be 4usec as i can set spectral_fft_period =1 then
>>> (fft_period+1)*4uS= 4uS. But then what would be the use of
>>> spectral_period variable?
>>>
>>>
>>> Kamran
>>>
>>> _______________________________________________
>>> ath9k-devel mailing list
>>> ath9k-devel@lists.ath9k.org
>>> https://lists.ath9k.org/mailman/listinfo/ath9k-devel
>>>
>>>
>>
>
_______________________________________________
ath9k-devel mailing list
ath9k-devel@lists.ath9k.org
https://lists.ath9k.org/mailman/listinfo/ath9k-devel
