动态演化环境下软件失效的在线预测关键技术研究

项目名称： 动态演化环境下软件失效的在线预测关键技术研究

项目编号： No.61202097

项目类型： 青年科学基金项目

立项/批准年度： 2013

项目学科： 计算机科学学科

项目作者： 张鹏程

作者单位： 河海大学

项目金额： 24万元

中文摘要： 在动态演化环境中，系统和环境的改变及人的误操作都可能会导致软件失效的发生，如何在线预测并预防软件失效的发生是当前亟待解决的重大问题。本项目给出了基于前瞻系统行为在线预测软件失效的新颖解决方案，主要包括四个方面。首先采用接口自动机，对环境、系统和人的行为进行建模，能够显式地表达系统对环境和人的不同假设。其次，用基于场景的规约来表示环境、系统和人交互过程中的关键属性，并定义其精确的运行时四值语义。再次，结合设计时模型和运行时信息，提出一种在线构建不久将来行为模型的方法，并面向基于场景的规约预警可能发生的失效，该方法能有效地控制模型构造的复杂度，避免了状态空间爆炸问题。最后，提出一组在线预防向导，通过调控系统、环境和人的行为，使得系统能及时避免失效的发生。本研究为在动态演化环境下软件失效的有效防止提供新思路，从而提高软件系统在部署以后的可靠性和安全性。

中文关键词： 软件失效；监控；在线预测；恢复；

英文摘要： In dynamic evolution environments, unsafe run-time changes of systems and environments, and human misuse may compromise the correct execution of the entire systems and lead to failures. How to predict and prevent the occurrence of software failures is a major issue to be solved in this research area. The project gives a solution based on looking ahead the behaviors of the systems that can predict failures in the future. It includes the following four aspects. Firstly, the project plans to use interface automata to model the behaviors of environments, system and human, so to clearly model the system assumptions for environments and human. Secondly, the project uses scenario-based notations to model the desired properties and defines its four-valued runtime semantics. Thirdly, the project proposes a new approach that by combining design-time and run-time techniques, constructs the on-line behavior model of near execution future, and check whether the model satisfies desired properties to predict future failures. The approach can efficiently control the complexity of model construction and avoid state explosion problem. The project also designs a set of prevention and recovery guides which can make the systems avoid failures to be occurred in advance. Overall the project provides new ideas and targets for effective

英文关键词： software failure；monitor；online prediction；recovery；