High-risk LEO Satellite Network Path Detection Based on Spatial and Temporal Delay Anomaly Analysis

Tianyu Zhang,Hewu Li,Jun Liu,Lu,Qian Wu,Shaowen Zheng,Zeqi Lai,Yuanjie Li

DOI: https://doi.org/10.23919/jcc.fa.2023-0077.202307

2023-01-01

China Communications

Abstract:The gradual deployment of Low-Earth Orbit (LEO) mega constellations with inter-satellite links (ISLs) promises ubiquitous, low-latency, and high-throughput satellite network services. However, networked LEO satellites with ISLs are also at risk of routing attacks such as hijacking. Existing defenses against route hijacking in terrestrial networks can hardly work for the LEO satellite network due to its high spatiotemporal dynamics. To deal with it, we propose RPD, a high-risk routing path detection method for LEO mega-constellation networks. RPD detects abnormal high-risk LEO network paths by checking the consistency between the path delay and the geographical distance. This is efficiently achieved by combining in-band measurements and out-of-band statistical processing to detect the anomaly of the clustering feature in the reference delay matrix. RPD avoids the recalculation of the header cryptographic marks when the handover occurs, thus greatly reducing the cost and improving the performance of high-risk path detection. Experiments showed that the proposed RPD mechanism achieves an average detection accuracy of 91.64% under normal network conditions, and maintain about 89% even when congestion occurs in multiple areas of the network and measurement noise is considered. In addition, RPD does not require any cryptographic operation on the intermediate node, only minimal communication cost with excellent scalability and deployability.