Performance optimization of Automatic Modulation Classification for different signal and channel types

The introduction of Cognitive Radio (CR) in mobile communication systems led to numerous opportunities to enhance spectral accessibility through Dynamic Spectrum Access (DSA). Automated Modulation Classification (AMC) confirms spectral white spaces and helps separating primary from secondary users. A significant number of different approaches have been presented and evaluated during the last years. Algorithms are generally developed and optimized for certain lengths of signal and transmission channels. The target for AMC optimization is to minimize the time interval required for classification and to make the system robust against different sources of distortion. The former coheres with the minimum signal length that is needed for the classifier to work, the latter can be simulated with different communication channel models. Therefore in this paper results are presented, which reflect the performance of AMC as a function of the signal length and for different modulation schemes and channel types.