Research On UAV Detection And Identification Based On Radio Frequency Fingerprint Technology Under Wi-Fi Signal Interference

UAV has been widely used in various fields.Since unauthorized UAVs intruding into the airspace have great security risks,especially in the combat of UAV clusters,fast and accurate identification of legal UAVs is particularly important.Radio frequency fingerprint identification is an emerging radio identification technology,has the advantages of anti-cloning attack and does not need to pre-load complex security algorithms,so it is of great significance to identify radio frequency fingerprints of UAV communication signals.However,since the frequency band of UAV communication signals is the same as that of some Wi-Fi signals,how to identify UAV communication signals in the presence of Wi-Fi signal interference has great practical application significance.Thesis mainly completes the following five aspects of work:1.Based on the frequency domain bandwidth characteristics of Wi-Fi signals and UAV communication signals,a scheme for filtering Wi-Fi signals in UAV communication signals is proposed.Under the premise that the UAV communication signal and the WiFi signal do not overlap in the frequency domain,for the mixed signal in the frequency domain,the maximum of the energy may be located in the communication signal of the UAV or in the Wi-Fi signal.Experiments were carried out to verify the effectiveness of the filtering scheme.2.A self-designed algorithm is proposed.After filtering out the Wi-Fi signal in the communication signal of the UAV,by extracting the time-domain envelope characteristics of the communication signal of the UAV.The extraction threshold of the key signal segment of the UAV communication signal can be adaptively determined,and extract the key signal segment of the UAV communication signal based on the threshold.3.Use the power spectral density method to extract the steady-state characteristics of the radio frequency fingerprint of the UAV communication signal.And based on four machine learning algorithms,the corresponding radio frequency fingerprint identification models were built respectively.4.Using 4 established RF fingerprint recognition models based on machine learning,respectively for the three cases of "without Wi-Fi signal interference","with Wi-Fi signal interference" and "after filtering Wi-Fi signal interference" of the UAV communication signal set for testing.Under different signal-to-noise ratios,the recognition rate results in the three cases are compared,and the effectiveness of the Wi-Fi signal filtering scheme proposed in thesis is verified.5.Aiming at the problem that it takes a long time to identify UAV communication signals,a method based on multi-resolution analysis is proposed to reduce the dimensionality of key signal segments of UAV communication signals and then use power spectral density to extract its RF fingerprint features.The effectiveness of the scheme is verified by comparing the time required to identify the UAV communication signal before and after dimensionality reduction.