function Sout = add_phase_noise( Sin, Fs, phase_noise_freq, phase_noise_power )
Oscillator Phase Noise Model
INPUT:
Sin - input COMPLEX signal
Fs - sampling frequency ( in Hz ) of Sin
phase_noise_freq - frequencies at which SSB Phase Noise is defined (offset from carrier in Hz)
phase_noise_power - SSB Phase Noise power ( in dBc/Hz )
OUTPUT:
Sout - output COMPLEX phase noised signal
NOTE:
Input signal should be complex
EXAMPLE ( How to use add_phase_noise ):
Assume SSB Phase Noise is specified as follows:
-------------------------------------------------------
| Offset From Carrier | Phase Noise |
-------------------------------------------------------
| 1 kHz | -84 dBc/Hz |
| 10 kHz | -100 dBc/Hz |
| 100 kHz | -96 dBc/Hz |
| 1 MHz | -109 dBc/Hz |
| 10 MHz | -122 dBc/Hz |
-------------------------------------------------------
Assume that we have 10000 samples of complex sinusoid of frequency 3 KHz
sampled at frequency 40MHz:
Fc = 3e3; % carrier frequency
Fs = 40e6; % sampling frequency
t = 0:9999;
S = exp(j*2*pi*Fc/Fs*t); % complex sinusoid
Then, to produse phase noised signal S1 from the original signal S run follows:
%%% first signal S
nfft = length(S);
res = fft(S,nfft)/nfft; % normalizing the fft
f = fs/2*linspace(0,1,nfft/2+1);% choosing correct frequency axes
res = res(1:nfft/2+1); % amplitude of fft(taking the half length of nfft)
figure, plot(f,abs(fftshift(res)));
%%% second signal S1
nfft = length(S1);
res = fft(S1,nfft)/nfft; % normalizing the fft
f = fs/2*linspace(0,1,nfft/2+1);% choosing correct frequency axes
res = res(1:nfft/2+1); % amplitude of fft(taking the half length of nfft)
figure, plot(f,abs(fftshift(res)));
I am eager to know, if the Fc is very high (e.g. 300MHz), does this code still work? I found a very strange waveform when I set Fc to a high frequency.
I have below OFDM model;
FFT size = 64
Subcarrier spacing = 1e5(Hz)
Symbol duration = 1/subcarrier space = 1e-5(s)
duration of 1 data sample = 1e-5/64 (s) (IFFT data is coverted to serial and become1/64)
in this case, "Fs" in this program should be 1e-5/64 (s) ? or any other values should be used ?
Very good. Thank you. Is there anyway to increase the precision of the approximation? This method is valid only for low phase noise values. How can one produce higher phase noise values?
However, I need to point out a minor bug in the code. I noticed that the phase noise I received from the code was higher than anticipated. Then I noticed on line 219 that the normalization is incorrect. Alex is correct in that (2*M-2) is needed to compensate for the inverse DFT; however this normalization (line 219) should be M vs. (2*M-2) since on line 222 he creates the two-sided spectrum by adding the negative frequency spectrum. In essence line 222 adds the rest of the normalization (M-2). Once I corrected this the phase noise I get is within 1 dB of anticipation vs. 6 dB.
Finally, there is a little mistake, which makes no difference in the result since the phase noise is generated via a random variable. However, to be numerically correct line 231 should have a -j vs. +j in the exponential.
Hi I wonder how this model can satisfy the real shape of the oscillator that firs we have a 1/f^3 and then 1/f^2 and then noise floor.
and it doesn't have close to carrier phase noise.
I don't know how based on which model you generate phase noise?
thanks
how come i run this program but it says Not enough input arguments. ?
Comment only
06 Oct 2008
Nick A
This approach is very elegant. However, it limits the amount of low frequency phase noise extracted from the PSD by the original sampling frequency. Any ideas about circumventing this limitation?
Thanks
Nick
18 May 2008
Agus Suhendar
i have problem, hope u can help me..
after qam mapping and ifft the signal is in time domain and complex form, with BW=7 MHz the problem is how to sampling the complex time domain signal with sampling factor=8/7 or Fs=8MHz
thanxs b4..
Comment only
20 Feb 2008
CT Lin
thank you
20 Feb 2008
chuntel Lin
thank you
Comment only
01 Aug 2007
john kedziora
Works very well!
19 Dec 2006
thomas hÃ¶hne
Great work. Easy to adapt to own needs. The plot of the generated PSD could be done better.