A measurement study regarding quality of service and its impact on multiplayer online games

Abstract

In this paper we study the impact of network parameters including delay, delay jitter, and packet loss on the outcome of multiplayer online games. We use a measurement methodology that allows for automating game experiments using an artificial intelligence that runs on the game clients and eliminates the need for human interaction. Thus, we were able to perform thousands of experiments with different network configurations to investigate their influence on the game outcome systematically. Complementary to previous studies, we provide a solid data basis for statistical analysis. We derive probabilistic models that allow for calculating score and win probabilities and verify their accuracy using our measurements. As a result we are able to predict the outcome statistics of an ideal game with respect to network conditions.