Evaluating the Latency of Clients by Player Behaviors in Client-Server Based Network Games

Network games are becoming more and more important in the last years. Latency on the Internet is a well-known problem for interactive network games. In client-server (C/S) based network games, players only know their own network delay. Though latency of other clients can be easily got from game servers, most of game servers do not provide these useless information for players because of no benefits to game providers. In network games, because of network delay, two players can hardly see each other at the same time, though they have the same range of visual scope. From the view of local client, the movement of a nearby player looks like to be made sooner or later, and so is the nearby player. So, reaction of the nearby player to the movement of the local player will occur at different time from local view, when he is under different network condition. We can evaluate the latency of clients by the distinction of reaction time. In this paper, we propose a method to evaluate latency of players from local client according to reaction of players and rules of the game system. To gain a better understanding of reaction of players, we analyze a trace of Crazy Tank, a modified 2D C/S based tank game. Then we use PopKart, a popular network game, to verify feasibility of our method in real network games. Simulation shows that our method achieves significant accuracy in latency evaluation.