Research On Intrusion Detection Model For Airborne Network Based On Deep Learning

In recent years,in order to provide passengers with better cabin entertainment services,more and more airlines have combined traditional in-flight entertainment system with Internet service,extending the convenience and diversity brought by Internet to the flight of high altitude.However,the increased connectivity of airborne network could also be an open door to cyber-attacks as their complexity keeps growing.With intrusion behaviors threatening flight safety,passenger privacy,and airline brand image,research on airborne network security is of great significance.Firstly,this paper analyzes current research status,comprehensively considers the characteristics of airborne network architecture,and proposes a modularly designed intrusion detection model based on following four steps: data acquisition module,data preprocessing module,data analysis module,and data post-treatment module.Secondly,to meet the high detection performance requirement of airborne network,a novel intrusion detection method based on deep learning theory is designed and implemented.Considering the excellent feature extraction capability of deep learning neural network and the distinguished classification performance of support vector machine,a hybrid detection method combining deep belief network and support vector machine is proposed.Finally,the intrusion detection simulation experiment is conducted to study the effect of network depth,kernel function and training data amount on detection results,and the parameters of proposed detection method are optimized based on the experimental results.For different kind of intrusion behaviors,a parallel detection method is proposed,and experiments are conducted to verify the detection performance of proposed method for different types of attacks.Compared with other detection methods,the superiority of proposed method in detection performance is verified.