基于正常行为聚类的卫星通信网异常检测方法

针对卫星通信网中可能出现的因地球站被捕获、盗用、伪造和非法用户入侵等严重威胁卫星通信网自身安全的问题,提出通过对地球站正常行为聚类的方法实现卫星通信网的异常检测.提出了一种改进的KFCM(komel fuzzy C-means)聚类算法,对KFCM算法初始聚类中的确定做了优化,对聚类个数的确定做了改进.同时,在数据的预处理方面提出了一种主要属性的选择方法.通过几个经典数据集和卫星地球站真实数据的实验证明,该方法具有较好的检测效果,对于卫星通信网异常检测具有较高的实用性.

Research on anomaly detection method of satellite communication network based on normal behavior clustering

Aiming at the pro blems ex ist ing in satell ite communicat ion netwo rk, which may seriously thr eaten it s secur ity such as secret leakag e and int rusion caused by lost o r fal sification o f portable satellite ear th statio n, an anomaly detection method of satellite communication netwo rk by cluster ing ear th station normal behavior w as put forward. A KFCM ( komel fuzzy C-means) algorithm was proposed w hich impro ved the init ial cluster center selection and decision of best part it ioning . Exper iment s on ty pical dataset s and the real log-data of the satellite communicat ion netw or k w ere carried out . T est r esult s show that the metho d is feasible and effect ive on satellite earth stat io n anomaly detection.