Towards statistical client prediction - analysis of user behavious in distributed interactive media

Distributed interactive media such as networked multiplayer computer games offer users the opportunity to interact and share experiences within a virtual environment. More often than not, these interactions are required to be performed in real-time, a constraint which poses problems given the underlying network capabilities used to transmit information. In these real-time distributed systems, the amount of information that needs to be shared between participants in order to maintain complete game-state fidelity is too large. As a result, trade-offs must be made over what information requirements are necessary to maintain a level of consistency that will provide adequate quality of interaction for the users. One possible solution to this problem is the use of statistical modeling techniques that attempt to capture the individual behaviour of system users. These models can then be used to predict the likely future behaviour for the users, thus reducing the shared information requirements. In this paper we present some preliminary analysis of the behaviour of users within a distributed interactive application, with a goal towards future work of attempting to develop and incorporate statistical models of user behaviour for the purpose described above