Traffic models for video game applications

Efficient resource utilization in access networks requires a characterization of the types of traffic the network will handle. This traffic characterization can be done in two steps. The first step involves the identification of the applications that use network resources and the percentage of time they are active. The second step describes the traffic generation pattern of an active application. In this work we analyze network traces generated by two popular action video game applications in order to characterize their traffic generation pattern. This characterization is achieved by obtaining a statistical description of different parameters inter-packet time, inter-burst time, burst-length, and packet-length associated with a traffic model that employs a state machine. The model is shown to be robust, that is, the parameters of the model can be characterized using the same probability distribution independent of the application. It is shown that the gamma (which encompasses the exponential distribution as a special case) distribution can be used to characterize the random parameters of the model. Observed values of the parameters are presented.