Modeling network traffic as images

The paper presents a network measurement approach to represent samples of network packet header data as frames or images. With such a formulation, a series of samples can be seen as a sequence of frames or video. This enables techniques from image processing and video compression to be applied to the analysis of packet header data to reveal interesting traffic properties. We show that traffic images can reveal sudden changes in traffic behavior or anomalies. Using a combination of visual modeling and trace-driven simulation, we evaluate how the design factors impact the representation of dynamic network traffic. In particular, we study the impact of sampling rate and retained DCT coefficients on the network traffic data representation.