Trust Evaluation Strategy for Single Sign-on Solution in Cloud

Design Of Trust Evaluation Model Of Node

In this paper, each cloud server node is regarded as a unified entity of SP (Service Provider) and IDP (Identity Provider) and as a peer node for single sign-on in cloud. The distributed and dynamic characteristics of the cloud resources determine that SP adopted the policy of “fully trust” or “totally do not trust” to the verification certificate provided by the IDP. As for the previous centralized single sign-on model, the credibility and determinacy will decrease with the increase of the number of the entities. Usually, for a certain peer node in the group which contains a large number of peer nodes often can’t obtain the whole information of the other peer nodes. Therefore, this paper proposed a trust evaluation model for the peer node N_i.

Each peer node N_i has a list that records trust evaluation value of the other peer nodes. The trust value of the peer node can be represented by triples T_tuple(α, β, γ), and 0 ≤ α, β, γ ≤ 1, α+β+γ=1. Here, α refers to the probability of “can trust” of the peer node; β means to the probability of “can’t trust” of the peer node; γ represents the probability of uncertainty of the peer node. According to the different understandings of the trust of the peer node (for example, peer node N_i considers it is trustable only when α > 0.9, while N_j thinks it is fully trustable as long as α > 0.7), the values of α, β and γ are continuous rather than discrete. According to the role and function, there’re four types of trust: trust of SP, trust of IDP, recommended trust of the SP and recommended trust of the IDP.

Suppose the trust evaluation value of source peer node N_s to destination peer node N_d is (α₁, β₁, γ₁) and trust evaluation value of the other peer nodes to N_d is (α₂, β₂, γ₂), the finally trust value of N_s to can be calculated through: