Towards understanding the relationship between game type and network traffic for cloud gaming

Cloud based games have significantly larger network bandwidth requirements when compared with traditional online games (up to two orders of magnitude). Differences in transmission characteristics can be observed not only between various cloud gaming platforms, but also between different games played on the same platform. As the differences between games on the same platform can be very high (up to several Mbit/s), we can witness significant impacts on game performance and consequently the end user Quality of Experience. In this paper, we present a first step in the attempt to quantify this phenomenon through measurements of 18 games which differ in type, perspective, and input characteristics running on an open source cloud gaming platform- GamingAnywhere. We analyse gathered video samples and calculate the values of spatial and temporal video metrics defined in the literature. We identify the relation between two sets of video metrics and compare them with the network traffic characteristics, primarily bandwidth usage and packet rate. We also annotate and analyse the impact of user input and the context of the game on the application level.