Automatic digital modulation identification is a technique for recognizing the modulation scheme of an intercepted signal. It is required in many applications, such as software radios. This work presents a new intelligent technique for identification of digital modulation. In this technique wavelet packet analysis is used to extract the prominent characteristics of received signal. It utilizes a support vector machine as the classifier to identify the type of modulation. Genetic algorithm has bee used to select the most suitable features and the best model of support vector machine. This technique is used to identification the considered modulations set. Evaluation results show this new technique has high accuracy even in low levels of signal to noise ratios. This high efficiency is achieved with only eight selected features and the best possible kernel parameter of classifier which have been optimized using genetic algorithm.

Keywords