Robust and Adaptive Credibility Model

Abstract

In reality, it is not unusual that a Trust Management Service (TMS) experiences malicious behaviors (e.g., collusion or Sybil attacks) from its users. Credibility-based trust management of cloud services poses cloud services protection issues because it is difficult to know how experienced a user is and from whom malicious behaviors are expected. On the one hand, the quality of trust feedbacks differs from one person to another, depending on how experienced s/he is. On the other hand, attackers can disadvantage a cloud service by giving multiple misleading feedbacks (i.e., collusion attacks) or by creating several accounts (i.e., Sybil attacks). Indeed, the detection of such malicious behaviors arises several challenges including: (i) Consumers Dynamism where new users join the cloud environment and old users leave around the clock which makes the detection of malicious behaviors (e.g., feedback collusion) a significant challenge, (ii) Multiplicity of Identities where users may have multiple accounts for a particular cloud service which makes it difficult to detect Sybil attacks because malicious users can use multiple identities to give misleading information, (iii) Attackers Behaviors where it is difficult to predict when malicious behaviors occur (i.e., strategic VS. occasional behaviors).