Re: [Discuss-gnuradio] OFDM Benchmark Change Modulation

[Ben Reynwar]

On Tue, Jan 4, 2011 at 6:49 AM, You Lizhao <address@hidden> wrote:
> Hi all,
>
> Recently I want to implement a OFDM based multi-rate system, and I am using
> benchmark_ofdm_tx/benchmark_ofdm_rx programs in ofdm directory. I know I can
> use unlock/lock mechnism to disconenct/connect the exising blocks to change
> modulation block, and it indeed works. However, I also notice that the
> difference between all modulation, i.e. BPSK, QPSK, QAM16, QAM64, is
> rotated_const paramter in gr.ofdm_mapper_bcv block. So I write a new block
> howto.ofdm_mapper_bcv which can change rotated_const  (d_constellation in
> gr_ofdm_mapper_bcv.h/.cc) with a defined function call. But it does not
> work. When I change BPSK to QPSK by altering rotated_const, I can receive
> nothing but TIMEOUT in QPSK demodulator. If I still use BPSK demodulator, it
> works in compromised performance with some packets failed to pass CRC check.
> Why does it happen, since I already change rotated_const? Is there any
> misunderstanding on this ofdm benchmark program?
>
> Furthermore, I also find that if I use QAM16, it seems that the packet
> reception simuation is very bad, about 50% PRR or even lower. When
> using QAM64, PRR~0%, almost all packets failed CRC check. I think if I use
> channel coding, the situation will be better. I am wondering how to add
> channel coding block to existing architecture. Is there any possible
> examples?
>
> Any comments are welcomed! Thank you very much!
>
> Regards,
> -Lizhao
> _______________________________________________
> Discuss-gnuradio mailing list
> address@hidden
> http://lists.gnu.org/mailman/listinfo/discuss-gnuradio
>
>

I don't know whether this is relevant or not but just in case:

The QAM constellations produced by blks2impl/qam.py aren't set up for
differential modulation which was causing me some problems with the
benchmark_qt_loopback2.py program.  In my github repo
(github.com/benreynwar/gnuradio) i've got a modified qam.py which
fixes this.

Cheers,
Ben