Re: [Discuss-gnuradio] pilot symbols in ofdm

On 11.06.2014 14:43, Karan Talasila wrote:
> Hi
>
> Looking at the ofdm receiver at ofdm_rx grc, I see that schmidl and cox
> synchronisation algorithm uses sync1 and sync2 which are two ofdm
> training symbols that help in frequency offset correction and also
> timing estimate to detect the start of the packet.
>
> Then there is channel estimator block in grc that again uses preamble
> ofdm symbol to detect coarse frequency offset. In the entire
> implementation I haven't seen the use of pilot symbols in each ofdm
> symbol for either channel estimation or synchronisation. So does that
> mean that if you are using a packet structure with a preamble there is
> no need of pilots for any receiver processing.
>
> I get this doubt because benchmark_rx.py which uses ofdm gives you a
> terminal option saying occupied tones which can be set by user at
> terminal. So if you can set or increase occupied tones to any number
> less than fft length, then that means you can be flexible with
> increasing or decreasing pilot symbols to a value that you want. Am I right?

Yes, you can use zero pilot carriers. They are used in the
ofdm_frame_equalizer_vcvc, but our current implementations aren't very
good. For longer frames, I recommend using pilot symbols (distributed in
time, not frequency!), but for short packets, the initial estimation is
sometimes enough, if your setup is two USRPs close to each other and you
have something like 10 OFDM symbols total.

M


_______________________________________________
Discuss-gnuradio mailing list
Discuss-gnuradio@gnu.org
https://lists.gnu.org/mailman/listinfo/discuss-gnuradio