Turbo decoding Lc Vs Eb/N0 and Coding gain

[kpojha]

Hi,
I am checking the performance of turbo decoder (1/3 rate) by doing simulation for different Eb/N0 and for BPSK modulation. I have few doubt about the entire simulation process.

1. I generated the AWGN noise (0-mean,1-variance) and adjusted the amplitude of the noise by multiplying.
Ec_No_dB = Eb_No_dB-10*log10(3.0); because of 1/3 rate
NoiseAmp = (1/sqrt(2.0))*(pow(10.0,-1*(Ec_No_dB/20.0)));
Noise = NoiseAmp*(randn(1,K)+j*randn(1,K));

Is this process is correct ????

My signal constellation point is +/-1+j*0;


2. Lc the SNR information needed by turbo decode

Ideally Lc = (Eb/N0)*(Rate/BW).
As I dont have the BW concept so
I am considering Lc = (4.0)*(1.0/3.0)*pow(10.0,((Eb_No_dB[siLoop])/10));
I have doubt on this expression? Any clarification

3. Ideally for BPSK simulation a BER of 1e-5 can be achieved at Eb/N0 of 9.1dB Because the turbo has a rate of 1/3 so the coding gain should be checked against a reference point (9.1 - 10*log10(3)). I could not see any gain from the above reference.

Could some one can explain on this ???


Thanks
Kiran Prakash