[srslte-users] Low Leve PHY sync method usage

[Sylvain Munaut]

Hi Andre,

> Yes, it would return the first sample after the PSS. Since the PSS is
> sent in the 7th OFDM symbol of the first and the sixth subframe, you get
> the start of the 8th symbol of either the first or the sixth subframe.

Ok, good to know what I observed is what should be happening :)


> By default we only receive a half frame because we can assume that we
> received at least one PSS (or at least a fraction of it).

Sure, I understand the rationale, it was just a bit confusing that
it's called "frame_size", but I know naming stuff is always the hard
part :p


> We usually initial sync and MIB decoding at 1.92 MHz sample rate but
> then switch to the actual bandwidth of the cell. So it works with full
> bandwidth too.

I can see how that would save on CPU usage, however not really an
option here. Even if I prototype on a USRP that can do that, the
hardware I will end up using wouldn't be able to change rate and keep
timestamp easily.


> I don't think you need it for the PSS sync if you pass the FFT size
> manually but as your signal is using 30.72 MHz, you need to make sure to
> also use the default LTE FFT sizes. Have you passed the CMake flag
> accordingly or have you set the static variable "bool use_standard_rates
> = true;"?

I modified the example code from synch_file to add

srslte_use_standard_symbol_size(true);

at the beginning, but as you said, I don't think this changes anything
for the PSS when you manually specify the FFT size.


> Maybe you can also share a minimal non-working example.

http://people.osmocom.org/tnt/lte/lte_be_796m_1536.cfile

This is an example capture. This is at 15.36Msps rather than
30.72Msps, but same difference, it's just a bit quicker to init.

My understanding is that :

./lib/examples/synch_file -s 1024 -l 76800 -i lte_be_796m_1536.cfile

should find the sync for it.

But it doesn't, and I don't understand why.


Cheers,

     Sylvain