Considering the delay characteristics and real-time requirements of encrypted WIFI traffic,a time-delay aware UAV operation pattern recognition method based on machine learning was proposed.Firstly,the proposed machine learning framework regards the encrypted data stream was taken as a time series,and the reweighted l1 norm regularization was used to set the feature se-lection and performance optimization into the expected total cost function,which integrated the misclassification/false detection and delay cost.In addition,in order to solve the uncertainty of packet arrival time when estimating the delay cost function,the max-imum likelihood estimation was used to obtain the packet arrival time.Finally,the performance of the proposed method is evaluat-ed by using the real WIFI data flow of private UAV,and the evaluation results show the effectiveness of the proposed method.
维普
万方数据