IMPROVING ANOMALY DETECTION IN BORDER GATEWAY PROTOCOL TIME-SERIES DATA

Abstract

The Internet infrastructure relies on the Border Gateway Protocol (BGP) to provide basic routing information. Where the abnormal routing events, resulting from direct or indirect anomalies, weaken the Internet connection and network balance. For this purpose, we have proposed several algorithms to improve the detection of anomalies in BGP time-series data. Our methods show that detected anomalies are more realistic and that the selected features are generally consistent across time-series. The performance evaluation is offered using various machine learning techniques.