Year: 2025
Type: conference-paper
Abstract: Due to its precision and efficiency, Automatic Dependent Surveillance-Broadcast (ADS-B) is extensively used in air traffic control systems to ensure smooth ATC operations. However, the protocol's inherent openness exposes it to significant risks from malicious cyber-attacks. The detection algorithms in previous studies have the problem of low attack detection rate, and for the identified attack behaviors, current studies rarely propose corresponding recovery methods. This study introduces an ADS-B attack detection and recovery framework utilizing an interclass multilevel orientation-extreme learning machine-generalized regression neural network (IMO-ELM-GRNN, IMOEG). Based on real flight data, we combine different types of attack vectors with ADS-B data, simulate and construct a variety of attack data sets, and use the interclass multilevel orientation-extreme learning machine (IMO-ELM) model to detect and mark the attack data. Finally, the generalized regression neural network (GRNN) model is used to recover the marked abnormal data. The approach presented in this paper enables the efficient and precise detection of anomalous targets carrying attack vectors in ADS-B data, and perform targeted recovery on the data points where the abnormal targets are located. We use multiple performance indicators to evaluate the detection and recovery performance of the model and compare it with other mainstream detection and recovery methods. The final experimental results show that the performance indicators of our detection and recovery model in the face of various attacks are better than other models and methods, with better overall detection effect and wider applicability.
Authors: Meng Yue, Tian Gao, Sunshuo Shi, Zhijun Wu
Institution: Civil Aviation University of China
Language: English
Subfield: Computer Networks and Communications
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
Awards: G1890901198, G272075324, G2827442430
OpenAlex ID: W4410813834