Study On Radio Frequency Fingerprint Identification Technologies Of The Wireless Communication Based On The Constellation Diagram

The security access and secure communication system technique that established by using the physical characteristics of wireless communication equipment and channels is called physical layer security technique.In recent years,the physical layer security technique is being studied by more and more scholars.The radio frequency fingerprint authentication system of wireless communication networks is a new type of security authentication system that can effectively authenticate wireless devices and users.Because the radio frequency fingerprint is the physical feature of the wireless communication device itself,the radio frequency fingerprint information established on the physical layer is not easy to be imitated or modified.Therefore,the communication security of wireless network can be protected from the bottom layer of wireless communication system.The physical layer security technique based on radio frequency fingerprints can effectively solve the problem of wireless network security authentication.This thesis explores,the methods of the radio frequency fingerprint extraction and identification authentication from the following aspects.1.The domestic and foreign research of the radio frequency fingerprinting technique is analyzed.Theoretical analysis and experimental verification are used to improve the existing radio frequency fingerprint extraction and identification scheme based on the constellation diagram.Aiming at obtaining clear and discrete constellation diagram at the receiving end,the thesis proposes an average minimum mean squared error channel estimation algorithm.The algorithm firstly constructs a new 802.11 n pilot frequency structure,then transmitter and receiver perform down-sampling and over-sampling processing.Finally,the frequency response of the channel is obtained via the known training sequence and pilot.When the signal to noise ratio is 20 d B,the simulation results indicate that the recognition accuracy of the system can reach 88.2%when the average minimum mean square error channel estimation algorithm is used.However,the recognition accuracy of the system can reach 85.6% when the eight pilot minimum mean square error channel estimation algorithm is used.The recognition accuracy of the system is 78.2% when the conventional minimum mean square error channel estimation algorithm is used,and the recognition accuracy of the system is only18.4% when no channel estimation algorithm is used.2.To deal with the problems of high computational complexity and long training time of the current radio frequency fingerprint classification and recognition algorithms,this thesis proposes a joint decision algorithm based on euclidean distance and amplitude distance for classification and recognition.After obtaining a clearer constellation diagram,the K-means clustering algorithm is used to get K clustering center points in the constellation diagram,and the K clustering center points are used as the radio frequency fingerprint of the device.The euclidean distance and the amplitude distance between the cluster center point of the input device and the cluster center point of the device in the fingerprint database are used as the decision threshold.The results indicate that compared with the one-dimensional classification and recognition algorithm,the recognition accuracy of the two-dimensional classification and recognition algorithm is greatly improved.The two-dimensional classification and recognition algorithm can increase the recognition accuracy by 7.8% at maximum while the signal to noise ratio is 5 d B.Compared with the K-nearest neighbor classification and recognition algorithm,the recognition accuracy of K-nearest neighbor classification and recognition algorithm is improved by about 1.9 d B when the recognition accuracy is 90%,but the computational complexity of the system.3.In order to verify the theoretical derivation and simulation analysis data,the thesis builds an experimental platform to verify the previous data.This thesis combines the universal software radio peripheral to perform the radio frequency fingerprint extraction and identification verification on actual wireless communication equipment.When the signal-to-noise ratio is 20 d B,10 groups of CC2530 Zig Bee devices are identified by the radio frequency fingerprint.Experimental results show that the radio frequency fingerprint identification technology based on constellation can achieve good results when the signal-to-noise ratio is greater than 20 d B.