Radio Frequency fingerprinting is the transmitters identification in wireless communication process, applying in wireless security, whose difficulty is the effective feature extraction. The feature extraction approaches for RF fingerprinting can broadly be divided into transient and steadystate analysis. Considering the difference of transient waveforms, we use transient intensity as the fingerprint feature to identity transmitters, which is defined as the ratio of signals’ mean power and peak magnitude. The classification performance of notebook wireless network adapts in different models and USB wireless network adapts in different series have been evaluated from experimental simulation, combined with features of the magnitude of HilbertHuang Transform and the Short Time Fourier Transform in timefrequency domain. It is demonstrated that the feature of transient intensity has some advantages, such as high accuracy and short time of classification.