Abstract:
The video game industry is very popular nowadays. According to a study, the approx imate worth of this industry is $159.3 billion with 2.7 billion gamers worldwide. As
video games are very popular and easily accessible, they can be used as a tool for pri vate communication to avoid online censorship. In the paper "A First look at Private
Communications in Video Games using Visual features" published by Wajid et al. it is
shown how one can use video games for secret communication and private messaging.
There are mainly two limitations in this paper. Firstly, the messages crafted in games
can be detected and read by anyone using manual human supervision. Secondly, if a
large dataset can be gathered, one can train a neural network and automatically detect
these messages crafted in games.In the proposed idea of this project, the results of the
above-mentioned paper were to improve by solving these two issues. As in video games,
one is allowed to make fine changes in cyberspace, using this property of video games
adversarial examples can be easily generated in this cyberspace and automatic detection
methods can be fooled. Also if messages are divided into different shares using cryp tography and then these cryptographic shares can be implemented in-game then even
with manual human supervision one cannot detect the messages. So, by using these
techniques the results are successfully improved in this thesis work.
