Re: [Discuss-gnuradio] BER Example - With QT Gui Chooser for Different Modulations Issue

[Cinaed Simson]

On 1/28/19 8:49 AM, Luis Felipe Albarracin Sanchez wrote:
> Hello Cinaed,
> 
> Thank you for the response,
> 
> I have detailed your Flowgraph, and I placed a "QT GUI Constellation
> Sink" after the chunks to symbols block, and what I found is that the
> modulation part only modulates in B-PSK, and this is beacuse you set it 
> in the "Chunks to symbol block" with a dimension of "1", once I placed
> different "Symbol Table" and a "dimension" of 2 in the "Chunk to symbol
> block" the correct modulation appeared, but the BER gave really high
> values (even if in the "Constellation Decoder" is set to  QPSK or DQPSK).

I already answered this - it's documented on the flowgraph - and I
answered it in a previous email.

BER only has 2 points - 0,1 - the constellation points aren't used.

The 8PSK decoder outputs 1,0 which creates an inverted pulse relative to
the reference input to the BER - the random number generator - and the
BER sky rockets.

So instead of using -1,1 in the chunks to symbols, I had to use 1,-1.

Is my flowgraph the correct way to calculate the BER? I have no idea.
The only thing I can say about it is the BER is 0 without white noise.

I've never done a BER calculation. When the BER is high on physical
MODEM I buy a new one.

The only thing I know is if you want to calculated bit error rate on the
order of ppm after adding white noise, and the computational error
without adding white noise is on the order percent, then the BER after
adding white noise would be impossible to detect.

And I know very little about QT - I just use the QT GUI blocks for
debugging.

But my intuition tells me you can't change the modulation object on the
fly while the flowgraph is running - but don't let that stop you from
trying.

> 
> I believe you only modulated in BPSK and in the decoder it matched the
> constellation points for -1 and 1, but not for any other constellation
> points, I would appreciate if you correct me if i am wrong.
> 
> Kind regards.
> 
> On Mon, Jan 28, 2019 at 5:12 AM Cinaed Simson <address@hidden
> <mailto:address@hidden>> wrote:
> 
>     On 1/27/19 7:03 PM, Luis Felipe Albarracin Sanchez wrote:
>     > Hello Cinaed,
>     >
>     > Thank you so much for all the work in the BER, i tested both of the
>     > flowgraphs and they are very accurate regarding the BER
>     measurements, i
>     > have a couple of questions if you don´t mind:
>     >
>     > 1- In the latest flowgraph you sent (my-ber-test.grc from Jan
>     25th) how
>     > do you change the modulations in the chunks to symbols block, in the
>     > flowgraph you sent me the flowgraph table was fixed with the values of
>     > "-1, 1", what line of commands should be place there to switch to
>     other
>     > modulations?
> 
>     Correct - I already explain why.
>     >
>     > 2- The BER measurements you sent me seem so accurate,
> 
>     As I already indicated, you still need to add a complex gaussian noise
>     source.
> 
>      I wonder if you
>     > could take a look at the flowgraph i sent you
> 
>     No, use the last flowgraph I sent you.
> 
> 
>     > (my_ber_simulation_LFA.grc), what i did there was connecting a QT GUI
>     > Chooser to the functions i thougth would change the constellation
>     > selected,  do you have any ideas why  my change is not working, when i
>     > change in the QT GUI Chooser?, i would appreciate your help
>     regarding this.
>     >
>     > 3- Attached is a flowgraph i have devloped for doing modulations up to
>     > 64 QAM, please take a look at it, and if you have any ideas on how i
>     > would make a good BER measurement of higher modulations (32 -QAM and
>     > 64-QAM) please let me know. To select the modulatios, the easiest
>     way is
>     > to define in the default value of the QT GUI chooser the value of
>     3 (for
>     > 32-QAM) and the value of 4 (for 64 QAM).
> 
>     I not going to look your flowgraph - you should be able to add the QAM
>     constellations to the last flowgraph I sent you.
> 
>     And to be clear, I don't believe can change the modulation of the
>     flowgraph with QT. Since you're the believer, the burden of proof is on
>     you.
> 
>     You need to first demonstrate you can add the 64QAM, and 32QAM to the
>     last flowgraph I sent you with DER 0.
> 
>     Then add whatever QT modifications you need to make QT change the
>     modulation and demonstrate the BER is 0.
> 
>     Or just demonstrate you can add QT code which changes the modulation to
>     last flowgraph I set you. If can't make the changes so it has BER of 0,
>     adding different of more modulations isn't going to help.
> 
>     -- Cinaed
> 
> 
>     >
>     >
>     > Once again thanks a lot for all the help.
>     >
>     > Kind regards.
>     >
>     > On Fri, Jan 25, 2019 at 3:20 PM Cinaed Simson
>     <address@hidden <mailto:address@hidden>
>     > <mailto:address@hidden <mailto:address@hidden>>>
>     wrote:
>     >
>     >     It bothered me that the random source wasn't working.
>     >
>     >     So I took at it again last night and discovered I needed to
>     unpack a bit
>     >     from the random source stream.
>     >
>     >     Using the random source, the following work:
>     >
>     >       BPSK
>     >       QPSK
>     >       8PSK
>     >       16QAM
>     >
>     >     using the pull down menu.
>     >
>     >     There is, however, one caveat: for 8PSK you have to enable the
>     complex
>     >     Multiply Const by -1 block after the Chunks to Symbols block.
>     >
>     >     Alternatively, you could change Chunks to Symbols from -1,1 to
>     1,-1 -
>     >     but that's many more key strokes then just enabling and
>     disabling the
>     >     Multiply Const block.
>     >
>     >     They all work with the GLFSR source too and with the same caveat.
>     >
>     >     And you were right you don't need the PSK symbol table - you
>     can use
>     >     Chunks to Symbols -1,1 (except for 8PKS.)
>     >
>     >     Also, I had to add a unpack k=1 block after the decoder -
>     otherwise for
>     >     DPSK, 8PSK and 16QAM there was a vertical offset and BER sky
>     rockets.
>     >
>     >     The unpack k=1 is not needed for BPSK and QPSK - but it
>     doesn't create
>     >     any problems so I left it enabled.
>     >
>     >     You just need to add the complex noise source
>     >
>     >     Have fun with QT.
>     >
>     >     Actually, there is already a QT menu - the properties of
>     Constellation
>     >     Object. You might want to look at the source - just figure
>     what you need
>     >     to do to keep it open.
>     >
>     >     -- Cinaed
>     >
>     >
>     >
>     > --
>     > Eng. Luis Felipe Albarracin
>     > Msc. Telematics / MBA
>     > PMP
>     > CCNA/CCDA/CCNP/CCDP/CCIP
>     > ITIL v3 Foundation
>     > "Die Grenzen meiner Sprache bedeuten die Grenzen meiner Welt"
> 
> 
> 
> -- 
> Eng. Luis Felipe Albarracin
> Msc. Telematics / MBA
> PMP
> CCNA/CCDA/CCNP/CCDP/CCIP
> ITIL v3 Foundation
> "Die Grenzen meiner Sprache bedeuten die Grenzen meiner Welt"