基于RFMI特征选择的卫星网络入侵检测技术

With the rapid advancement of satellite internet technologies and the maturation of sophisticated constellation deployments and networking capabilities, cyberattacks targeting satellite nodes have escalated in both frequency and sophistication. Space-based assets now serve as prime targets for malicious actors seeking to either exfiltrate critical data or exhaust onboard computational resources through coordinated exploitation attempts. In this context, intrusion detection systems have emerged as critical components of space network security architectures, serving as vital defense mechanisms against evolving vulnerabilities in space-based network ecosystems. Satellite systems require dedicated security architectures due to their geographically isolated deployment, inherent resource-constrained nature, and the challenges posed by multidimensional data complexities.To address these challenges, this study proposes a resource-efficient intrusion detection methodology that employs optimized feature selection. We introduce a novel RFMI-based algorithm designed to systematically identify optimal feature subsets through tripartite optimization: maximizing feature importance, strengthening class correlation, and minimizing redundancy. Experimental results demonstrate that the proposed algorithm exhibits high robustness, achieves strong performance across evaluation metrics, and adapts effectively to the satellite internet environment.