A Novel Hierarchical Method for Digital Signal Type Classification

Abstract:- Automatic signals type classification (ASTC) is a technique for recognizing the modulation scheme of an intercepted signal and has seen increasing demand in both military and civilian. Most of previous tech-niques can identify only a few kinds of signal types. Also, usually, they request high SNR to achieve a mini-mum acceptable performance and don’t include higher order and new types of signals (e.g. QAM types).The work presented here proposes novel high performance method for identification of digital signal types. In this method it is proposed a hierarchical multiclass classifier based on support vector machine (SVM). The inputs of this classifier are higher order moments and cumulants (up to eight). Genetic algorithm (GA) is used to select the best model of SVMs in order to improve the performance. Simulation results show that proposed method can recognize a lot of digital signal types and achieves a highly probability of classification even at low signal to noise ratios (SNRs)