基于贝叶斯网络的网构软件可信性评估模型

对于运行在开放、动态、难控的互联网环境的网构软件,其可信性评估是一个重要课题,但目前大量研究中可信性计算多是基于黑盒的,没有深入考虑系统结构,且评价指标过于单一.因此,提出了一个基于贝叶斯网络的网构软件可信性评估模型.该模型通过对网构软件进行结构分析,根据其结构模式,建立多层的网构软件可信性评估指标体系.基于贝叶斯网络采用自底向上逐层分析计算的方法,对网构软件的各组成实体及其系统整体的多方面可信性指标进行评估,形成统一的可信性结果,并使用客观数据对其进行修正.实验证明,该模型可以明确、客观地对网构软件的可信性进行评估,并能够对网构软件的设计、开发和部署提供帮助.

A Dependability Evaluation Model for Internetware Based on Bayesian Network

Dependability evaluation is an essential issue for Internetware which is in open,dynamic and decentralized Internet environment.But dependability is generally evaluated by simple metrics based on black box without considering system structure in many cases.In this paper,an evaluation model of dependability for Internetware based on Bayesian network is proposed.It builds a multitier dependability evaluation metrics system by structure analysis and structure pattern of Internetware.It uses bottom-up approach on the basis of Bayesian network to calculate dependability metrics of Internetware system and its components from many aspects.A unique dependability metrics is obtained and amended by objective data as a result.Experimental results show that the model can evaluate dependability of Internetware clearly and objectively.It also contributes to design,development and deployment of Internetware.