Dear sharmin

The noise power is given by the variance of the noise, so when you generate a noise, generate a normal one with variance one and to control the power of the noise, you have juste to multiply by a standerd deviation.

BR

Hi Sharmin, a close estimation, the expected noise power, is  E[x2]= SIGMA2( i.e. variance of the noise), and we can desirable noise power by varying SIGMA.  

To generate a noise with a desired power (variance), you should start with the target SNR of the system (in dB). Suppose that the signal is normalized so its variance is 1. Then select the standard deviation (sigma) of the noise as:

sigma=10^(-SNRdB/20);

where SNRdB is the desired Signal to Noise Ratio in dB (typically from 0 to 60).

After that, scale your previously generated AWGN noise by sigma:

noise=sigma*noise;

This works for both real and complex valued noises. Hope it helps.

You should define what you mean by "noise power". If you are adding white noise to a signal (in Matlab) you can simply do  signal = signal + meann+randn(1,1)*variancee, where

signal is your input signal

meann is the mean of the Gaussian noise (https://en.wikipedia.org/wiki/Gaussian_function)

and variancee is the variance (note wrong spelling is use so as to avoid conflict with Matlab built in functions).

The above command will add a scalar noise to each element of your signal vector.

Depends on what you are using to simulate the noise. I agree with the above answers and adding another formula for Band-limited white noise block in Simulink.

http://www.mathworks.com/help/simulink/slref/bandlimitedwhitenoise.html

thank you very much to all of u respected researchers.

it was really helpful.