Modulation Classification based on Statistical Moments

Abstract

The ability to identify the modulation of an arbitrary signal is desirable for a number of reasons, including signal confirmation, interference identification, and the selection of proper demodulators. A modulation classification algorithm using statistical pattern recognition techniques has been developed and tested on numerically simulated signals. This algorithm uses statistical moments of both the demodulated signal and the signal spectrum as the modulation identifying parameters. The basis for the classification routine is a set of formulated probability distributions which were developed by generating and statistically analyzing a large set of numerically simulated signals. The resulting classification equations were tested on an independent set of numerically simulated signals.