Generating Realistic Data for Network Analytics

In this paper, we provide a novel approach that uses a generative adversarial network to produce synthetic network traffic. The intent is to leverage this synthetic data to improve the robustness of machine learning algorithms that perform analysis on communication networks. In our experimental results, we demonstrate that a generative adversarial network can construct samples of network traffic that are statistically similar to an original set of reference samples. Additionally, we provide insight into the performance of our approach when evaluating different varieties of generative adversarial networks for their ability to produce and converge to realistic output.