Byzantine resistant reputation-based trust management

Abstract

Cloud computing is very useful for improving distributed applications performance. However, it is difficult to manage risks related to trust when collaborating with unknown and potentially malicious peers. Besides, trust evaluation is the target of dishonest behaviors trying to disturb the control process. In this paper, reputation-based trust management models for cloud computing are proposed. These Peer-to-Peer (P2P) reputation models are based on the interaction between peers. Using evaluations and feedbacks, a central entity can estimate the trust of a given peer. Three approaches are proposed to estimate the trust: PerronTrust, CredTrust and CredTrust-trust. They are studied, simulated and compared between them and to two existing methods for trust under several attack scenarios. Our analysis clearly shows that the third approach CredTrust-trust combining the concepts of trust and credibility in an appropriate way is the most efficient to avoid malicious behaviors and to guide and advise future executions in the open cloud in term of selecting the dependable and reliable peers in cloud environment.