Big network traffic data visualization

Visualization is an important tool for capturing the network activities. Effective
visualization allows people to gain insights into the data information and discovery of
communication patterns of network flows. Such information may be difficult for human to
perceive its relationships due to its numeric nature such as time, packet size, inter-packet
time, and many other statistical features. Many existing work fail to provide an effective
visualization method for big network traffic data. This work proposes a novel and effective
method for visualizing network traffic data with statistical features of high dimensions.
We combine Principal Component Analysis (PCA) and Mutidimensional Scaling (MDS) to
effectively reduce dimensionality and use colormap for enhance visual quality for human
beings. We obtain high quality images on a real-world network traffic dataset named ‘ISP’.
Comparing with the popular t-SNE method, our visualization method is more flexible and
scalable for plotting network traffic data which may require to preserve multi-dimensional