i need to recognize the modulation type of incoming signal in IF(intermediate band) .that is, i want recognize the modulation of signal is FSK or non FSK(such as PSK or QAM or ASK)?
how can i design this classifier?
remark:
the non FSK modulation have pulse shape.
i test several methods but all of them have problem and work in a special condition.
the methods have been tested : wavelet, zerocrossing, cumulant, ...
please help me.
Why don't you just detect the phase change ?
I mean...lets assume you know frequency of transmitted signal and assume all other necessary data...u know phase(freq shift) shift...
Do FFT with standard Cross-correlation method...
Wavelet thing is very suitable for Non-stationary signals...and here I am assuming that u are playing with stationary signal...
As my experience Quasi-likelihood methods are exelent for MPSK modulation classification. there is the capability to classify also pi/4-shifted QPSK. It has some problems also, such as it needs a fine carrier frequency estimation and food timing recovery.