|
Disincentivizing
Malicious Users in RecDroid Using Bayesian Game
Model Bahman Rashidi+
and Carol Fung Virginia Commonwealth University,
Richmond, Virginia, USA Abstract RecDroid is an Android smartphone permission control framework which provides fine-grained permission control regarding smartphone resources and recommends the permission control decisions from savvy users to inexperienced (novice) users. However, malicious users, such as dummy users created by malicious app owners, may attempt to provide untruthful responses in order to mislead the recommendation system. Although a sybil detection function can be used to detect and remove some dummy users, undetected dummy users may still be able to mislead RecDroid framework. Therefore, it is not sufficient to depend on sybil detection techniques. In this work, we investigate this problem from a game-theoretical perspective to analyze the interaction between users and RecDroid system using a static Bayesian game-theoretical formulation. In the game, both players choose the best response strategy to minimize their loss in the interactions. We analyze the game model and find both pure strategy Nash equilibrium and mixed strategy Nash equilibrium under different scenarios. Finally, we discuss the impact from several parameters of the designed game on the outcomes, and analyzed the strategy on how to disincentivize attackers through corresponding game design. Keywords: Game Theory, Android, Recommendation
System, RecDroid +: Corresponding author: Bahman Rashidi Journal of Internet Services and Information Security (JISIS), 5(2): 33-46, May 2015 [pdf] |