连续时间马尔科夫链模型上的模型检测算法研究与工具实现

ISCAS OpenIR > 基础软件与系统重点实验室

连续时间马尔科夫链模型上的模型检测算法研究与工具实现

高杨

Major

计算机软件与理论

Supervisor

詹乃军

2013-05-27

Degree Grantor

中国科学院大学

Degree Level

硕士

Place of Degree Grantor

北京

Keyword

模型检测连续时间马尔科夫链形式化概率模型检测工具

Abstract

连续时间马尔科夫链（CTMC）在网络性能分析、模型检测和系统生物学等领域受到了广泛的关注。本文关注以连续时间马尔科夫链为模型、以条件连续随机逻辑（CCSL）为性质描述语言的概率模型检测问题。CCSL通过引入条件概率运算符扩展了Aziz等人提出的连续随机逻辑（CSL），从而可以表达和检测CTMC更多的性质，本文针对CCSL逻辑提出了基于参数乘积CTMC的近似模型检测算法，并分析了该算法的理论时间复杂度。进而，本文介绍了针对CSL和CCSL逻辑开发的概率模型检测工具CCMC已有的概率模型检测工具如PRISM和MRMC仅仅支持CSL二元Until公式，CCMC是第一个支持CSL全部语法形式和其扩展的概率模型检测工具。最后，本文通过一些实例分析展示了工具的性能和实用价值。;

Continuous-time Markov chains (CTMCs) have received considerable attentions in network performance analysis, model checking, and system biology. In this paper, we consider the model-checking problem of continuous-time Markov chains with respect to Conditional Continuous Stochastic Logic (CCSL). CCSL extends the logic CSL introduced by Aziz et al. with a conditional probabilistic operator, which allows us to express a richer class of properties for CTMCs. Based on a parameterized product construction, we propose an approximate model checking algorithm with complexity analysis. After that, we present CCMC (Conditional CSL Model Checker), a model checker for CTMCs with respect to formulas specified in CCSL. Existing CTMC model checkers such as PRISM or MRMC handle only binary CSL until path formulas. CCMC is the first tool that supports algorithms for analyzing nested until path formulas. Moreover, CCMC supports conditional CSL formulas. Some case studies are given showing the efficiency of the tool.

Subject

计算机软件 ; 软件理论

Language

英语

Content Type

学位论文

URI

http://ir.iscas.ac.cn/handle/311060/14845

Collection

基础软件与系统重点实验室

Recommended Citation
GB/T 7714

高杨. 连续时间马尔科夫链模型上的模型检测算法研究与工具实现[D]. 北京. 中国科学院大学,2013.

Files in This Item:
File Name/Size	DocType	Version	Access	License
gymain.pdf（1331KB）			开放获取	License	Application Full Text

If you have any objections to this item, please fill out the form below and the administrator will contact you as soon as possible.
Content:
Email：	*
Institution:
Verification Code:	Refresh

Any comments and suggestions are welcomed.
Title:	*
Content:
Email：	*
Verification Code:	Refresh

中国科学院软件研究所机构知识库