rng and randn on matlab

Hi there,

Can we say this code has a good approximation about the channel coefficient for each SNR?

rng(1)

SNR=10:5:20;

N_frame=100000;

taps=4;

for i=1:length(SNR)

for 1:N_frame

chan_coef=sqrt(1/2)*(randn(1,taps)+1i*randn(1,taps));

...

end

end

My result showed that in one SNR condition, the chan_coef will be extremely small, but didn't appear in other SNR condition

[Ashu]

Can you give more clarification on what results you are getting?

Answer 1

[vidyesh]

Hi mingcheng,

Since channel coeffecients are random in nature, their values can be extremely small at time. Also please note that the channel coeffecients you are generating are independent of SNR.

I would advice you to:

a) Generate a large numer of channel coeffecients, equal to the number of samples/symbols and the utilize those in your simulation.

b) Generate the channel coeffecients outside the loop iterating over different SNR values. As a result the channel coeffecients will be the same for all values of SNR and it will allow you to see the effects of increasing SNR clearly.

Refer to the below page for more information on how channel coeffecients can be generated.

Fading Channels - MATLAB & Simulink (mathworks.com)