Automatic modulation recognition of digitally modulated signals

A modulation recognizer that automatically reports modulation types of constant-envelope modulated signals is developed using zero-crossing techniques. The zero-crossing sampler, as a signal conditioner, has the advantage of providing accurate phase transition information over a wide dynamic frequency range. Signal parameters, such as zero-crossing variance, carrier-to-noise ratio (CNR), and carrier frequency, are estimated. Phase-difference and zero-crossing-interval histograms play the roles of features for modulation recognition. The classifier performance is given in the form of a confusion matrix. The obtained simulation results demonstrate that reasonable average probability of correct classification is achievable at CNR=15 dB and higher.<
>