Airborne network intrusion detection method based on deep learning

Abstract

The invention discloses an airborne network intrusion detection method based on deep learning. The method includes the steps of data packet monitoring, data packet collection, time stamp labeling, filtering sampling, feature mapping, character-type feature digitization, normalization processing, feature learning, performance evaluation and the like. Compared with traditional detection models, an airborne network intrusion detection model generated by the method provided by the invention has better detection performance, and decreases probability that an airborne network is attacked due to missed detection or false reporting.

Description

technical field [0001] The invention belongs to the technical field of network security, in particular to an airborne network intrusion detection method based on deep learning. Background technique [0002] Airborne networking has achieved rapid development worldwide and is expected to create a market worth $130 billion in the next 20 years. But the increased openness of an aircraft once considered the last cyber island has also opened the door to sophisticated and varied cyberattacks. The security issues brought about by openness pose challenges to the airborne network in the rapid development stage, and the research on airborne network security issues is imminent. [0003] Based on the concept of multiple independent security layers (MILS), Jacob proposed intermediate services such as labeling, filtering, and information flow control, and used a partitioned communication system to be responsible for the communication between MILS nodes. Laarouchi et al. proposed a soluti...

AI Technical Summary

Problems solved by technology

[0004] However, as network intrusion data is becoming more and more complex and feature-diversified, and limited by the constraints of time and space complexity, traditional machine learning algorithms often perform poorly and are prone to "dimension explosion". It leads to high false positive rate, high false negative rate and poor adaptive ability of the detection model

Images