Cyber range exercises are important to validate security tools and train personnel to enhance their testing skills. The background web traffic for cyber range exercises must
emulate a real network. The currently available traffic generator tools can only replay the captured network traces, or generate simple packet streams and do not consider the representative end-users behaviour. Generating realistic and useful web traffic
incorporating a varsity of user behaviour models is a challenging
research task.
In this work, we present a web traffic generator based on Markov model, Dirichlet distribution, and Hybrid distribution. We evaluate our models using a large dataset containing one million web sessions, and our results show that the entropy of
output data and cross entropy is approximately same as the entropy of training data. Our tool has interactive GUI and supports various OS environments, and hence, can be used for various web traffic generation applications including CR.