Re: Reed-Solomon and AWGN

Antepani Mario wrote:
Please, excuse me, I have made some confusion.
I try to better explain:
My system is composed by a Bernouilli Binary generator with a
generation rate of Rate_gen=300Kbps*11/15. (sample time of uncoded
bit is 1/Rate_gen).
After the Reed Solomon (15,11) the bit rate is Rate_RS=300Kbps
(sample time of RS coded signal is 1/Rate_RS).
Then, a QPSK modulator and an AWGN.
After that, a QPSK demodulator, a Reed-Solomon Decoder and a BER
meter.
The fact is that we found two parallel BER curves:
putting in AWGN a sample time of 1/Rate_gen we found a worse BER
curve than putting in AWGN a sample time of 1/Rate_RS.

Yes, that's exactly what you should see. That's why I said if you are only interested in the shape of the curve, it makes no difference.

The "Symbol Period" you specify for the AWGN channel is only used to calculate the noise power it is going to generate; it does not have to be the same as the "sample time" of the Simulink signal going into it.
Typically, for performance curves, you specify Eb/No in terms of the uncoded bit energy, not the coded bits (it would be very difficult to compare different codes if you used the coded bit energy). See the documentation for the AWGN channel block to get a better idea of how it computes noise power based on your settings. Basically, the larger your symbol period, the larger your noise variance/power will be (all other options being constant).

When you use 1/Rate_gen, you are using the uncoded bit energy (this is probably what you want if you are comparing your results against some theoretical curves or other codes).

Idin Motedayen-Aval



First, you have to be careful here not to mix up sampling frequency
and bit/symbol periods. Saying your sample time is 300Kbps*15/11
is a
bit
ambiguous to me. Do you mean you're specifying "300e3*15/11" as
your
sample time or "1/(300e3*15/11)"? (I hope you're doing the latter).

Assuming your bits have a *sample time* of tb=1/(300e3*15/11), each
bit
will have a period of tb, and each symbol a period of ts=2*tb.
Thus for
your AWGN channel block you would specify symbol period as "2*tb"
or
"1/(2*300e3*15/11)".

But I think your question is one step deeper than this; ts above is
the
period of an uncoded symbol. The period of a coded symbol would be

"ts*11/15". Your settings are using the energy of a coded bit, and
your
colleague's settings use the energy of an uncoded bit.

So it depends on what you are trying to accomplish. If you want to

compare the performance of this system to an uncoded system, to be
fair,
you would set Eb/No using the energy of an uncoded bit (so you use
the
same amount of energy for each source bit). If you just want to
study
the behavior of the BER curve for this system by itself, then it
doesn't
really matter which number you use; the shape of the BER curve will
be
the same.

HTH,
Idin Motedayen-Aval

.

Relevant Pages

  • Re: Future cars will be 100% electric cars
    ... the on-line power use curve will begin to look more and more like ... with large-scale investments in large capacity energy storage ... would receive power on demand from space via phase conjugate laser ...
    (sci.energy.hydrogen)
  • Re: help please
    ... is energy along a curve minimum ... even then kindly let me know of the formula for energy along a curve ... The best material model of a cat is another, or preferably the same, cat. ...
    (comp.soft-sys.matlab)
  • Re: Dumb moments in tech in ST novels
    ... where energy goes into linear momentum and if ... Straighten it, and you fly away from the sun, curve it, and you fall into the sun. ... As you get closer to the sun, the amount of natural curvature increases to the point where you can't summon up enough energy to curve it any more. ...
    (rec.arts.startrek.tech)
  • Re: sources of gravity
    ... >Space-time should curve up to a point in space because gravity would ... >representing the space around us which is filled with vacuum energy) ... >small indeed if vacuum energy exists at the density quantum mechanics ... imagine parallel mirrors with a photon trapped ...
    (sci.physics.research)