hi everyone,

I am trying to simulate a OFDM system with AWGN channel, 4QAM symbol in
matlab.

Theoritically, the curve for BER should be the same as for BPSK

BER = Q(sqrt(2*Eb/No));

But I have a problem: When I normalized the symbols 1/sqrt(2)(1+j) (For
example), that doesn't work. But when I leave out the factor 1/sqrt(r), it works
perfectly. Anybody knows the answer? Please help.

Thank you very much

Kenny

# BER for 4QAM with AWGN Channel

Started by ●February 13, 2008

Reply by ●February 14, 20082008-02-14

My guess is that your noise power is off by the same factor of 1/sqrt(2).

Check how you generate complex noise and that each of the real and imaginary

part should have noise variance half of the total noise power.

Nandan

On Feb 13, 2008 6:10 AM, loc truong wrote:

> hi everyone,

>

> I am trying to simulate a OFDM system with AWGN channel, 4QAM symbol in

> matlab.

> Theoritically, the curve for BER should be the same as for BPSK

>

> BER = Q(sqrt(2*Eb/No));

>

> But I have a problem: When I normalized the symbols 1/sqrt(2)(1+j) (For

> example), that doesn't work. But when I leave out the factor 1/sqrt(r), it

> works perfectly. Anybody knows the answer? Please help.

>

> Thank you very much

>

> Kenny

>

Check how you generate complex noise and that each of the real and imaginary

part should have noise variance half of the total noise power.

Nandan

On Feb 13, 2008 6:10 AM, loc truong wrote:

> hi everyone,

>

> I am trying to simulate a OFDM system with AWGN channel, 4QAM symbol in

> matlab.

> Theoritically, the curve for BER should be the same as for BPSK

>

> BER = Q(sqrt(2*Eb/No));

>

> But I have a problem: When I normalized the symbols 1/sqrt(2)(1+j) (For

> example), that doesn't work. But when I leave out the factor 1/sqrt(r), it

> works perfectly. Anybody knows the answer? Please help.

>

> Thank you very much

>

> Kenny

>

Reply by ●February 14, 20082008-02-14

hello,

thanks for all your responses,

the white noise I did as below

sigma = sqrt(1/(2*Eb_No_linear));

AWGN_Noise = sigma.*(randn(1,N)+j*randn(1,N));

but when I simulate for Rayleigh Fading channel , with the same Noise, I need to scale the 4QAM symbols with factor 1/sqrt(2), then it works.

thanks,

Kenny

thanks for all your responses,

the white noise I did as below

sigma = sqrt(1/(2*Eb_No_linear));

AWGN_Noise = sigma.*(randn(1,N)+j*randn(1,N));

but when I simulate for Rayleigh Fading channel , with the same Noise, I need to scale the 4QAM symbols with factor 1/sqrt(2), then it works.

thanks,

Kenny

Reply by ●February 15, 20082008-02-15

I think what it means is that your average channel power (averaged over the

fading statistics) is not 1.

Nandan

On Thu, Feb 14, 2008 at 4:46 AM, loc truong wrote:

>

> but when I simulate for Rayleigh Fading channel , with the same Noise, I

> need to scale the 4QAM symbols with factor 1/sqrt(2), then it works.

>

> thanks,

>

> Kenny

>

fading statistics) is not 1.

Nandan

On Thu, Feb 14, 2008 at 4:46 AM, loc truong wrote:

>

> but when I simulate for Rayleigh Fading channel , with the same Noise, I

> need to scale the 4QAM symbols with factor 1/sqrt(2), then it works.

>

> thanks,

>

> Kenny

>

Reply by ●February 25, 20082008-02-25