Automatic modulation classification based on statistical features and Support Vector Machine

Automatic classification of the modulation format of a detected signal plays an important role in cognitive radio system and several military and civilian applications. Without previous knowledge of the received data, automatic modulation classification (AMC) becomes a difficult task. This paper discusses AMC of digital schemes which are commonly used in today´s communication systems. Several statistical features are extracted to represent signals and Support Vector Machines (SVMs) are then applied to classify the unknown modulation schemes. Experiment results demonstrate that the proposed AMC algorithm is a practical algorithm and it has high robustness in a wide range of SNR.