SCS Undergraduate Thesis Topics
|Justin Weisz||Srini Seshan||Detecting Cheaters in a Distributed Multiplayer Game Environment|
Cheating is currently a major problem in today's multiplayer games, and the most popular cheat used involves having the client software render information which is not in the current game view. In a first person perspective game, this type of cheating would allow a player to see their opponents through walls. Currently, many people are studying how this type of cheating can be detected and prevented in the context of client-server multiplayer games, but no one is studying how this type of cheating can be detected or prevented in a distributed context.
We are studying cheating behaviors in the context of a distributed publish-subscribe system. In this system, players create publications which describe where they are and what their actions are. Players also create subscriptions which register their location-based interests for other players in their vicinity. The first type of cheating involves a player who publishes themselves at multiple locations in the game world (i.e. "cloning"), and the second type of cheating involves a player who subscribes to more information than is currently in their field of view. We believe that both of these types of cheating can be detected by using a machine learning algorithm to classify patterns of behavior based on previous publication and subscription histories.