Wireless Network Security Enhancement Based On Radio Frequency Fingerprint

With the rapid development of various Internet of Things applications,more and more wireless devices are connected to the network,which brings great convenience to people's lives and certain security risks at the same time due to the openness of the wireless network.Traditional identity authentication mechanisms based on cryptographic protocols face greater information security risks in the Internet of Things scenarios with simple device structures and limited computing capabilities.The communication parties in the Internet of Things have asymmetry in storage computing power,power supply endurance,business flow,etc.,so that terminal devices in the downlink direction cannot use complex identity recognition algorithms to implement authentication of network center nodes.Therefore,the inherent characteristics of radio frequency fingerprints and the reciprocity existing between the radio frequency fingerprints of the communication parties are discussed in this thesis.Furthermore,to enhance the security of Io T access,a suitable method for identifying the communication devices in the Internet of Things scenarios,which uses radio frequency fingerprint technology to achieve the bidirectional identification of wireless Io T devices are proposed.Different from the existing researches that only focus on the inherent characteristics of the individual RF fingerprints of the transmitter,the underlying mechanism of the RF fingerprints of the transceivers participating in the communication by analyzing the generation mechanism of the RF fingerprints is talked.First,based on the theoretical basis of the reciprocity of the wireless channel,the rule of reciprocity of RF fingerprints between devices on both sides of the communication is studied.By building a Software Definition Radio experiment platform,Bluetooth signal transmission and reception in a real wireless environment are achieved,and the reciprocal characteristics of the RF fingerprint based on the differential constellation trajectory are verified.Second,an RF fingerprint reciprocal transformation network based on Autoencoder is proposed.The network could produce the RF fingerprint of one communication party by transferring the known RF fingerprint of another party,so that the RF fingerprint acquired in the uplink can be used in the downlink.The experiment results show that the RF fingerprint reconstructed by the transformation network has the same accuracy performance on identity recognition compare to the real RF fingerprint,which proves that the proposed reciprocal transformation network in this thesis could learn the reciprocal characteristics of the RF fingerprints of communication parties in a high performance.The generated fingerprint is able to characterize the real fingerprint.On this basis,a bidirectional identification method for the Internet of Things devices that uses the rule of reciprocity of radio frequency fingerprints is projected.Specifically,in the uplink direction,the Internet of Things central node uses the idea of abnormality detection to check whether the identity of the Internet of Things terminal device is authorized,to determine whether illegal users are exist.Furthermore,based on the reciprocity rule of RF fingerprints,the Internet of Things central node uses the RF fingerprint of the terminal device which is obtained from the uplink direction to predict the fingerprint of the central node seen by the terminal device in the downlink direction.In this way,the training of central node identity recognition network by the Internet of Things terminal can be migrated to the central node,so as to reduce the energy consumption of the Internet of Things terminal.The trained identification network by the central node is sent to the terminal of the Internet of Things,then the terminal could use the observed downlink signal to infer whether the identity of the central node is authorized.Experiment results show that the bidirectional identification algorithm proposed in this thesis could realize the identity recognition of communication parties.