开放分布系统安全中的Bayes信任模型

为了解决开放分布系统的安全问题,本文提出了一个Bayes信任模型.借鉴人类的信任概念,该模型根据直接经验和推荐信息可以计算出一个实体对其他实体的信任值.在该模型中,把实体间交互成功的概率作为信任的重要指标,选择Beta分布作为其先验分布,通过对实体间4种不同关系的分析,获得相应的成功交互概率的估计.最终可形成实体的相对稳定的交互系统,为是否和别的实体交互提供依据.该模型考虑了影响信任的主观因素和客观因素,且具有统计基础,为解决开放分布系统的安全问题提供了新方法.

A Bayes Trust Model for Open Distributed System Security

This paper proposes a Bayes trust model that employs the human notion of trust to address security issues in open distributed systems. Firstly, the successful interaction probability between two entities is taken as an important index of one entity's trusting in another, and Beta distribution is selected to be its prior distribution. Then, the relationships between two entities are classified into 4 kinds according to what if there exist intermediate entity and/or direct interactions, and the estimators of the successful interaction probability are analyzed in 4 situations using Bayesian approach. Finally, one entity's relatively fixed cooperation system is obtained. This cooperation system consists of all of the potential partners of this entity. This model considers not only subjective factors but also objective factors, and is based on statistical theory. It provides a new valuable method for open distributed system security.