Research On UAV Intrusion Detection Method Using Wi-Fi Traffic Features

Unmanned aerial vehicle（UAV）has been widely used in many fields,such as military target detection,agriculture,forest fire-prevention,and so on.At the same time,it also brings a series of security problems.The repeated occurrence of unauthorized UAV intrusion has caused great hidden dangers to national security and people’s security.At present,UAV intrusion detection mechanisms based on radar,vision,acoustics,and radio-frequency are easily interfered with by factors such as obstacles,light,noise,and equipment.The UAV intrusion detection method based on Wi-Fi data traffic is less affected by these factors.Therefore,this paper uses the Wi-Fi data traffic identification method to realize the rapid detection of UAVs.In view of the shortcomings of the current UAV intrusion detection model using Wi-Fi data traffic in detection accuracy,reducing sample size,and model interpretability,this paper proposes a new detection algorithm to study UAV intrusion.This paper deeply studies the method of UAV intrusion detection using Wi-Fi data traffic.The main research contents of this paper include:（1）A UAV intrusion detection method based on belief rule base（BRB）is proposed,which solves the problems of insufficient detection accuracy and an unexplainable model in the current detection algorithm of UAV intrusion using Wi-Fi data flow.BRB can effectively use various types of information to establish any nonlinear relationship between model input and output.This relationship is used to model and simulate any nonlinear model and optimize the model parameters,that is,implement interpretable machine learning.Here,the interpretability of BRB is derived from the transparency of confidence rules in the method and the interpretability of Extended Bayesian probability theory reasoning.And high accuracy can be obtained through a small amount of data.The experimental results show that after obtaining appropriate parameters and sufficient training,the accuracy and accuracy of the model are high.（2）To further improve the accuracy of the above UAV intrusion detection methods based on BRB,Wi-Fi traffic data are extracted to produce more indicators.When using ordinary BRB to build a model,too many indicators will lead to the explosion of BRB rule combination and affect the performance of model detection.To solve the problems of model combination explosion,Evidential reasoning（ER）algorithm is proposed for multi-attribute fusion,and then the fusion results are input into BRB.The best detection results are obtained.To reduce the influence of the uncertainty of initial parameters on the detection accuracy of the model,the optimization algorithm of the projection covariance matrix adaptation evolution strategy（P-CMA-ES）with projection operation is introduced into the model.A UAV intrusion detection method based on EBRB is further proposed.The experimental results show that on the basis of feature extraction of Wi-Fi data traffic,BRB algorithm combined with ER multi-attribute fusion algorithm further improves the accuracy of UAV intrusion detection and can meet the accuracy requirements of model detection.（3）Based on the proposed EBRB model,a UAV intrusion detection prototype system using Wi-Fi traffic characteristics is designed and implemented.The system includes system setting,data exchange,parameter setting,system testing,and other functional modules.This paper gives the system design,implementation,and test process,and describes the function and relationship of each module.