New results on a two-stage novel modulation classification technique for cognitive radio applications

We report here on new results for a novel modulation classification technique designed for a real-time cognitive radio (CR) system. The goal is a technique that is robust and efficient with a processing time overhead low enough to allow a CR to maintain its real-time operating objectives. We investigated classification of linear single-carrier modulations as well as multi-carrier modulations. The method uses the waveform's I-Q diagrams, and by employing clustering algorithms on them, determines the type of modulation being transmitted. For classifying single-/multi-carrier modulations, we further study existing methods to find and modify the appropriate Gaussianity test to distinguish between single-carrier signals and multi-carrier signals. This new modulation classification method is capable of determining the type of modulation scheme among various PAM, QAM, and PSK modulations. Its performance in classifying these modulations is comparable to the best available published classifiers. It is also capable of extracting features of OFDM modulations and can be further expanded to include any new modulation scheme with unique I-Q diagram.