A stochastic gamer's model for on-line games

Abstract

Online games are interactive competitions by players who challenge in a virtual environment. “Gamers Private Network (GPN®)”, developed by WTFast is a game network that provides client/server solutions to makes online games faster despite wide-area networks. “Response time, low latency and predictability are key features to GPN® success” [1]. To analyze the experiments we use stochastic models to investigate latencies, jittering and other delay behaviours in delivering packets between clients and game servers. Our previous analytical work used simple statistics and some Markov-models of the latency time-series [1]. They gave us black-box indications on the interaction between game parameters and latency in GPN® or non-GPN® setups. In this research paper we describe a more advanced stochastic model to connect internal random effects in the system (client, network, server) to its latency time-series. The mathematical description has previously been applied to bioelectrical signal processing in biological experiments, a happy but also realistic metaphor for game networks where players' reflexes are a key ingredient. This is another small step towards rational, quantified and service-oriented minimization/stabilization of GPN® latency.