Question about ofdm_sync_long block of gr-ieee802-11 module

GNU Radio
1

I am looking gr-ieee802-11 module.

a part of the source code is pasted here (
https://gist.github.com/gsongsong/3df4553fe96e2d62b83e)

I have a question on the following statement:

if(rel >= 0 && (rel < 128 || ((rel - 128) % 80) > 15)) in case COPY
if(((d_count + o) % 64) == 0) in case RESET

I am wonder what do these numbers(128, 80, 15, 64) and modulo operation
have logical meaning in IEEE 802.11p PHY frame?

I’ve read 802.11 OFDM specification, thus I have knowledge about it,
but I have no idea what are roles of these things in implementation.

Regards,
Jeon.

2

Hi Jeon,

in such a case, linking to the original file [1] including line number
might be more helpful, as it gives more context.

So, looking at this, it’s taking the delayed input samples and copies
them (offset-corrected) to the output, considering parts of the samples,
x = ignore, o = copy, starting at the frame start:

oooooooo|xx?oooooo|xx?oooooo|…
128 |16?128-16|16?128-16|…

I’m not too familiar with 802.11p frames, but that looks like Basti’s
leaving out some pilots in between the data, probably only copying the
preamble and data payload. But I could be wrong.

Greetings,
Marcus

[1]
https://github.com/bastibl/gr-ieee802-11/blob/master/lib/ofdm_sync_long.cc#L128

3

Hi,

On 09 May 2015, at 13:23, Marcus M. [email protected] wrote:

So, looking at this, it’s taking the delayed input samples and copies them
(offset-corrected) to the output, considering parts of the samples,
x = ignore, o = copy, starting at the frame start:

oooooooo|xxoooooo|xxoooooo|…
128 |16128-16|16128-16|

Exactly, its to remove the cyclic prefix (remove the first 16 samples of
every 80 samples). Unfortunately, the first 128 samples (the long
training sequence) dont fit into the same scheme. Therefore the special
case here.

Best,
Bastian

but I have no idea what are roles of these things in implementation.

Discuss-gnuradio mailing list
[email protected]
Discuss-gnuradio Info Page

–
Dipl.-Inform. Bastian B.
Distributed Embedded Systems Group
University of Paderborn, Germany

4

Dear Bastian,

I missed one another thing.

if(((d_count + o) % 64) == 0) statement in
https://github.com/bastibl/gr-ieee802-11/blob/master/lib/ofdm_sync_long.cc#L141

Is this because of a length of an OFDM signal being a multiple of 64,
which
is the number of subcarriers? (Except cyclic prefix)

Regards,
Jeon.

2015-05-09 22:32 GMT+09:00 Jeon [email protected]:

5

Dear Marcus and Bastian,

Thank you for your detailed answers.

Also, thank you for useful usage of gist. :slight_smile:

Regards,
Jeon.

2015-05-09 21:42 GMT+09:00 Bastian B. [email protected]:

6

Hi,

On 10 May 2015, at 04:21, Jeon [email protected] wrote:
I missed one another thing.

if(((d_count + o) % 64) == 0) statement in
https://github.com/bastibl/gr-ieee802-11/blob/master/lib/ofdm_sync_long.cc#L141

Is this because of a length of an OFDM signal being a multiple of 64, which is
the number of subcarriers? (Except cyclic prefix)

Yes, the next step is the FFT that works on blocks of 64 samples. The
line you quoted is executed when a new frame is about to start. It’s to
ensure that the new frame starts at a multiple of 64 samples.

Best,
Bastian