GMSK Bit error rate plot

I want to plot GMSK BER vs Eb/N0, but It gives me wrong result. do you know why?

hMod = comm.GMSKModulator('BitInput', true, 'InitialPhaseOffset',0);
hDemod = comm.GMSKDemodulator('BitOutput', true, ...
                    'InitialPhaseOffset',0);
hmod.bandwidthtimeproduct=0.3;
hdemod.bandwidthtimeproduct=0.3;
hError = comm.ErrorRate('ReceiveDelay', hDemod.TracebackDepth);

for EbNo=0:20;
hAWGN = comm.AWGNChannel('NoiseMethod', ...
    'Signal to noise ratio (Eb/No)','EbNo',EbNo);
      for counter = 1:100
      hAWGN.EbNo=EbNo;
      data = randi([0 1],300,1);
      modSignal = step(hMod, data);
      noisySignal = step(hAWGN, modSignal);
      receivedData = step(hDemod, noisySignal);
      errorStats = step(hError, data, receivedData);
      ber(EbNo+1)=errorStats(1);
      end
end
plot(ber);
semilogy(ber)
title('BER of GMSK Modulation');
xlabel('E_b/N_0 (dB)'); 
ylabel('Bit Error Rate');

[Mah Mhata]

all of the books and articles show another result.

[daniele scozia]

why when i use the object " errorRate " without the tracebackDepth the results are wrong?

[Stefano Marsili]

ok, a bit late but anyhow ...

I see 2 errors:

ber(EbNo+1)=errorStats(1);

here you want to build the average over the 100 packets but if you do it like that you just overwrite everytime the result with the ber of the last 300 bit packet

EbNo=0:20 .. check the energy of the signal. I saw there is and oversampling by 8 in modSignal, maybe you need to scale the EbNo accordingly

Hope it helps