DCNN-RFF-NFC: a novel design of NFC security using deep Convolution neural network-based RF fingerprinting

DCNN-RFF-NFC: a novel design of NFC security using deep Convolution neural network-based RF fingerprinting

연구 분야: Networking

학회: Neural Computing and Applications

To avoid tag misuse, especially desire Near-Field Communication (NFC) tag authentication. The most current fingerprinting approaches employ the physical-layer signal, which embeds the tag hardware faults for authentication. Nevertheless, the poor scalability of current NFC fingerprinting techniques for a large number of tags or their incompatibility with NFC protocols impedes the practical implementation of NFC authentication systems. Radiofrequency (RF) fingerprinting verifies by using distinct and random distortions in a received signal that are dictated by the transmitter’s attributes. Here, introduce a new deep learning-based RF fingerprinting system authentication technique. Although the accuracy of radio fingerprinting has been increased by the use of neural networks, there are two main causes behind this research. First, for pre-processing over radiofrequency signals, there are not many useful parameter possibilities. Second, in the case of crossing days, radio fingerprinting based on deep learning performs badly concerning temporal variations. Deep convolutional neural network designs can recognize devices in a broad range of realistic circumstances, including channel shifts, noise levels, training data quantities, and processing overheads, using both raw and processed IQ samples as input. Using an RF signal that corresponds to a one-bit transmission, the RF properties of the NFC tag are extracted to facilitate fast identification. Convolutional, recurrent, and fully connected layer-based neural networks are the three types of deep neural networks that are employed. Here, confirm through testing that deep learning-based algorithms are capable of uniquely identifying up to 98% of NFC tags.