EAGER: Bio-inspired Assurance and Regression Testing to Secure Organic Programs

软件工程需要解决一类日益重要的自适应和自我修复程序。这些程序感知环境的变化，并通过修改配置、库或程序代码做出反应。此外，程序修复和程序移植的自动化方法直接更改程序的源代码以修复、优化或添加新功能。总之，自我修改可以在发生变化时提供持续可用性，并可以强化系统以抵御入侵者。虽然自我修改的这种有机性质是一个强大的范例，但此类程序的整体可靠性和安全性面临风险。一旦发生任何调整，之前为推断其功能、安全性而开发的测试结果或保证案例就无效。然而，运行完整的回归测试周期或为保证案例开发新的证据和论据所需的时间可能需要数小时、数天或数周。该提案探讨了在适应过程中对有机程序及其相关测试套件和保证案例所做的更改进行动态和有效推理的可行性。鉴于这些程序模仿生命系统，并且许多适应已经通过进化算法实现，因此拟议的工作从大自然中汲取灵感，并使用生物启发技术来实现其目标。考虑到在线、不受监控的系统中漏洞的严重性，传统的功能故障和安全漏洞都是该提案的目标。更具体地说，该提案将首先探索确定保证的演变时间表的技术，找到必须进行重新验证的时间间隔。其次，它将模拟有机程序中的交互和测试覆盖范围。最后，它将使用受生物学启发的算法来选择、排序和生成回归测试的测试。该提案中进行的研究结果将为能够随着时间的推移验证有机程序的正确性和安全性奠定基础。 更广泛的影响包括共同指导本科生、跨学科团队参加国际基因工程机器（iGEM）竞赛，并建立保证案例来推理其项目的安全性。

Software Engineering needs to address an increasingly significant class of programs that are self-adaptive and self-healing. These programs sense changes to their environment and react by modifying configurations, libraries or program code. Furthermore, automated approaches for program repair and program transplantation change a program?s source code directly to fix, optimize or add new functionality. Together, self-modification provides continual availability in the presence of change and can harden a system against intruders. While this organic nature of self-modification is a powerful paradigm, the overall dependability and security of such programs is at risk. Prior test results or assurance cases developed to reason about their functionality, safety and security are invalid as soon as any adaptation occurs. Yet the time needed to run a full regression testing cycle, or to develop new evidence and arguments for an assurance case can take hours, days or weeks.This proposal explores the feasibility of reasoning dynamically and efficiently about the changes made to organic programs and their associated test suites and assurance cases during adaptation. Given that these programs mimic living systems, and many adaptations are already implemented with evolutionary algorithms, the proposed work draws inspiration from nature and uses bio-inspired techniques to achieve its goals. Both traditional functional faults as well as security vulnerabilities are targeted in this proposal, given the criticality of vulnerabilities in an online, unmonitored system. More specifically the proposal will first explore techniques to identify the evolution timeline for assurance, finding intervals at which re-validation must occur. Second, it will model interactions and test coverage in organic programs. Finally, it will use biologically inspired algorithms to select, order and generate tests for regression testing.The results of the research performed in this proposal will form the foundations for being able to validate the correctness and safety of organic programs over time. The broader impacts include co-mentoring undergraduate, inter-disciplinary teams to compete in the International Genetically Engineered Machine (iGEM) competition, and to build assurance cases for reasoning about the safety of their projects.