Post by paul005 on Mar 12, 2004 10:47:54 GMT -5
Hi,
I read your tutorial 1 with great interest.
As we know SNR=10*log10(power of signal/power of noise)
I have a signal which is based on a CW pulse pulse=sin(2*pi*(f)*t); %cwpulse. By doing some massaging and taking a Hilbert transform for the complex envelope we get the final signal which I call sig. Now to get the power of sig which will be better method?-
mean(sig)^2 or std(sig)^2. After we get the signal power it will be easy to determine the noise power to achieve that particular SNR. Once we get the noise power, do we add it to the signal to get the noisy signal? Which is correct-
noisy signal=sig+noise power or noisy signal=sig+sqrt(noise power).
Now if we like the noise to be a white gaussian noise, how can we go about using
randn(1,length(sig)) with the noise power calculated.
Also this signal is not in dB; to get it in dB we do:
sigdB=20*log10(sig)
I read your tutorial 1 with great interest.
As we know SNR=10*log10(power of signal/power of noise)
I have a signal which is based on a CW pulse pulse=sin(2*pi*(f)*t); %cwpulse. By doing some massaging and taking a Hilbert transform for the complex envelope we get the final signal which I call sig. Now to get the power of sig which will be better method?-
mean(sig)^2 or std(sig)^2. After we get the signal power it will be easy to determine the noise power to achieve that particular SNR. Once we get the noise power, do we add it to the signal to get the noisy signal? Which is correct-
noisy signal=sig+noise power or noisy signal=sig+sqrt(noise power).
Now if we like the noise to be a white gaussian noise, how can we go about using
randn(1,length(sig)) with the noise power calculated.
Also this signal is not in dB; to get it in dB we do:
sigdB=20*log10(sig)