Honest vs cheating bots in PATROL-based real-time strategy MMOGs

Abstract

Massively multiplayer online games (MMOGs) are being increasingly successful, since they allow players to explore huge virtual worlds and to interact in many different ways, either cooperating or competing. To support the implementation of ultra-scalable real-time strategy MMOGs, we are developing a middleware, called PATROL, that is based on a structured peer-to-peer overlay scheme. Among other features, PATROL provides AI-based modules to detect cheating attempts, that the decentralized communication infrastructure may favor. In this work we illustrate how we implemented honest and cheating autonomous players (bots). In particular, we show how honest bots can detect cheating bots in real-time, using strategies based on neural networks.