RF Fingerprinting techniques aim to authenticate a wireless emitter by the imperfections due to these components. It can be useful for authentication and network management for the future IoT networks. Various methods has been proposed using hand-crafted features and classic machine learning but nowadays many researchers try to apply deep learning architectures for RF Fingeprinting. Our contribution is based on Siamese Network, a deep learning architecture widely used by the face recognition community. We use the deep learning architectures proposed by the RF Fingeprinring community which processes the I/Q (In-phase and Quadrature) signal and the siamese network learning paradigms developed for the facial recognition to propose siamese architectures for RF Fingerprinting. One of the main advantage of the siamese network is the possibility to use one-shot learning and its ability to require a few data for the final implementation of the network. In this paper, we explain our implementation, our results and discuss about the potential benefits of our approach for final implementation in a wireless network.