In cyclostationary signal detection for particular type of signal modulation how to determine the value of cyclic frequency alpha so that it would give required detection.
Each modulated signal has a cyclic frequency such as BPSK ALPHA=2 it means we will have 2 picks frequency. So it depends on your modulated signal. It is why we call it feature detection, it means we must know alpha.
cyclic frequency is the symbol rate of a modulated signal (aka baud rate). In implementation, your frequency is the ratio between symbol rate and your sampling rate. If you take a look at the theory and understand why those baud rate peaks appear, you would know where you should expect them with your sampling rate.
When talking about "frequency detection", Fourier analysis in some or the other way is probably involved. If you take the Fourier transform of a cyclostationary signal, you should see a decomposition of the different frequencies of that signal. If your process is cyclostationary at different scales (say it has a daily cycle, a weekly cycle and a yearly cycle (road traffic density, for instance)) you would see high power on the frequencies corresponding to those cycles in your Fourier spectrum.
If the shape of your signal is complex, you can resort to the autocorrelation function, and take its Fourier transform. Then, again, the largest components should highlight at which frequencies your process is cyclostationary.
If I take fourier transform of my signal and to find its highest value if i plot it i would find the value of continous frequency where it is large but alpha is discrete frequency of fourier series. Please clarify.