Wavelet-Based Unwanted Traffic Time Series Analysis

Identifying traffic anomalies precisely and instantaneously is critical to network stability. Most studies have focused on analyzing unwanted traffic from a Darknet system. However, conventional methods of detecting anomalous activities from these data are not applicable to detection. We apply discrete wavelet transform (DWT) techniques for traffic signal decomposition and examine unknown anomalous activities from unwanted traffic data. Our work focuses on three unwanted traffic packets: TCP SYNs, TCP SYN/ACKs, and UDP packets and on three intervals: 10-ms, 100-ms and 1-s. Furthermore, we discuss the features of this approach and consider some of its possible realizations. Our goal is to reveal properties when wavelet techniques are used to detect network anormalies behavior.