Time-series Modelling of Server to Client IP Packet Length in First Person Shooter Games

Abstract

Modelling traffic generated by Internet based multiplayer computer games has attracted a great deal of attention in the past few years. In part this has been driven by a need to simulate correctly the network impact of highly interactive online game genres such as the first person shooter (FPS). Time-series models are important elements in the creation of realistic traffic generators for network simulators such as ns-2 and OMNeT++ as they account for the correlation between packets. In this paper we show that the time-series behaviour of FPS server-to-client packet lengths is well modelled by ARMA(1,1) processes. We report on data from six popular FPS games of the past 10 years including Half-Life, Half-Life Counter-Strike, Half-Life 2, Half-Life 2 Counter-Strike, Quake III Arena and Wolfenstein Enemy Territory. For each of these games we analyse sessions each comprising 2 to 9 players. In all cases ARMA(1,1) is an effective model. We also show that AR models of order 1 and higher fail to capture the packet size variance as effectively as ARMA(1,1). Finally we show that higher order ARMA models are no more effective in describing the time-series behaviour than the simpler ARMA(1,1) models.