Automatic identification of digital signal types is of interest for both civilian and military applications. Most of methods can only identify a few kinds of digital signal and usually need high levels of signal to noise ratio (SNR). This paper presents an efficient signal type identifier (ESTI) that includes a variety of digital signals. In this method, a combination of higher order moments (HOM) and higher order cumulants (HOC) are proposed as the features. A multi-layer perceptron neural network with resilient back propagation learning algorithm is proposed to determine the membership of received signal. We apply the genetic algorithm for feature selection in order to reduction of computational complexity. Simulation results show that the proposed method has high performance to identify the different types of digital signal even at low levels of SNR. This high performance is achieved with only seven features selected using genetic algorithm

Keywords