Abstract-Specific emitter identification (SEI) is the technique which identifies the individual emitter based on the RF fingerprint of signal. Most existing SEI techniques based on the transient RF fingerprint are sensitive to noise and need different variables for transient detection and RF fingerprint extraction. This paper proposes a novel SEI technique for the common digital modulation signals, which is robust to Gaussian noise and can avoid the problem that different variables are needed for transient detection and RF fingerprint extraction. This makes the technique more practical. The technique works based on the signal's energy trajectory acquired by the fourth order cumulants. A relative smoothness measure detector is used to detect the starting point and endpoint of the transient signal. The polynomial fitting coefficients of the energy trajectory and transient duration form the RF fingerprint. The principal component analysis (PCA) technique is used to reduce the feature vector's dimension, and a support vector machine (SVM) classifier is used for classification. The signals captured from eight mobile phones are used to test the performance of the technique, and the experimental results demonstrate that it has good performance even at low SNR levels.