cyclostationary signal

Started by June 25, 2007
Hello all,

I am trying to implement the algorithm found in the paper "Carrier
Frequency and Chip Rate Estimation Based on Cyclic Spectral Density of
MPSK Signals" by Z. Zhang et. al. IEEE

Basically, the algorithm states that if I calculate the cyclic
autocorrelation of the signal and take its fourier transform and plot
in in 3D (x-axis is chip rate, y-axis is the frequency, z-asix is the
PSD). Now I get a beautiful 3D graph but its all sorts of wrong. The
problem is that if I have a chip rate of 5MHz, i end up with the
second peak (first one being at the center frequency with chip rate
frequency = 0) around 6.2MHz. If my chip rate is 10MHz, the peak is
close to 10 but not exactly at 10! Similarly 12,16,18 etc. are off of
the axis. I am sure I am doing something ridiculous. Has anyone looked
at this paper or used this kind of a technique to estimate the Fc and
chip rate?

Thank you.