A game-theoretic analysis of deception over social networks using fake avatars

Abstract

In this paper, we formulate a deception game in networks in which the defender deploys a fake avatar for identification of the compromised internal user. We utilize signaling game to study the strategy of the deployed fake avatar when she interacts with external users. We consider a situation where the fake avatar as the defender is uncertain about the type of a connected external user, which can be a normal user or an attacker. We attempt to help the defender in selecting her best strategy, which is alerting to the system for detecting an attack or not alert. For this purpose, we analyze the game for finding the Perfect Bayesian Nash equilibria. Our analysis determines for which probability of the external user being an attacker, the defender should launch a defending mechanism.