Automated bot player detection for games using runtime verification

Abstract

In an online environment, players feel discouraged to play a game knowing that there are other players using software that automates their character. Bot detection for online games has not been a major interest area in software companies. The challenge lies in how one can detect an automated bot is being used. Bots degrade player experience because players who are putting their effort into a game are getting beaten by an automated character. The bots are programmed to gather materials repeatedly without resting, giving an advantage over human players who eventually rest. Some bots are programmed to have fast reflexes which can kill other players easily without letting the human player to react. This upsets players who wish to play the game fair. Detecting bots is not easy and may require code changes in various parts of the game. Monitor oriented programming gives way of extending functionality without altering the game code and creates a framework for the formal specification language and the game together, forming a system. In this dissertation we investigate the use of runtime verification as a method of checking for an automated bot inside a game. Monitor oriented programming is used to analyse the players' behaviour during a running game session and compares it with an ideal human behaviour, taking necessary actions when a violation occurs without altering the code. PolyLarva is the event-based runtime-verification tool developed in University of Malta used to enhance Doom Legacy with bot detection, thus shows how monitoring oriented programming can be used.