While radio fingerprinting has shown its efficiency in detecting rogue base stations and cloned wireless devices, the techniques proposed in this context focused on analyzing the turn-on transient portion of the signal. The proposed parameters do not allow building an accurate fingerprint. In addition, the detection of the transient requires that the transition between noise and transient does not occur gradually. We propose in this work a new signal fingerprinting approach based on discrete wavelet transformer. Instead of being extracted from the transient part, the signal fingerprint is computed on a part of the signal related to the transmission of a definite sequence of bits. This sequence is expected to be transmitted by every base station antenna in the network. To detect rogue base stations, we integrated the proposed technique to an existing security architecture called CODERA.