VisVerND: Visual Verification of Network Traffic Dataset

Abstract

Captured network traffic data is generally used to evaluate, validate and analyze novel solutions for various network problems. Hence, the quality and the completeness of the network traffic dataset are important for the analysis to yield correct results. However, the quality of the network traffic dataset is never questioned. In this work, we propose a systematic methodology to visually verify the quality and completeness of a network traffic dataset. The methodology analyzes the dataset for topological and protocol distribution properties and verifies whether it conforms to an ideal traffic dataset. It is observed that visual verification process helps to identify flawed and incomplete network traffic datasets.