Estimating Attack Risk of Network Activities in Temporal Domain: A Wavelet Transform Approach

Abstract

Analyzing network traffic data to detect suspicious network activities requires tremendous efforts because of continuously changing network traffic patterns and intrusion scenarios. Numerous research has been devoted to the task of identifying network anomalies while maintaining excellent performances. However, most studies focus on identifying network attacks without considering their temporal domain. Time information is useful for discovering patterns in network activities and understanding the changes in network traffic over time. This paper introduces an approach to discover network traffic patterns with time series analysis to estimate the level of attack risks. Classification is performed with machine learning techniques to assess the estimated attack risks. Findings from this study can increase the capability to detect network intrusions by analyzing the behaviors of temporal data and estimating their attack risks.