Automated Support for Cyber-Physical Systems Design: from Theory to Practice

背景。我们的社会依赖于网络物理系统（CPS）。CPS是软件密集型系统，广泛地与其物理环境交互。我们的大多数行业，包括汽车、能源和医疗保健，都依赖于CPS进行日常运营。例如，CPS监视和控制我们的汽车、发电厂和医疗设备。根据2020年行业研究CPS市场报告，2019年全球CPS市场收入为524亿美元，到2025年将达到896亿美元。尽管最近进行了投资和改进，但CPS设计仍然很复杂，容易出错，因此成本很高。例如，美国国防部估计，2020年F-35战斗机的Block4升级成本将达到144亿美元，该升级旨在提高飞机应对新出现威胁的能力。此外，CPS故障也可能是灾难性的。例如，在2018年至2019年期间，波音737 MAX的缺陷导致346人死亡。研究的挑战。工程师需要为CPS设计提供自动化支持。研究界在开发支持CPS设计的自动化技术方面投入了大量精力。尽管取得了许多成功，但仍有许多挑战阻碍了这些技术在实践中的广泛使用，特别是在工业模型中。本研究计划考虑了四个研究挑战：（N1）对搜索和检测CPS设计缺陷的自动化技术的需求，（N2）对理解问题原因的帮助的需求，（N3）对自动将人类工件（如CPS需求）转换为机器可处理输入的程序的需求，以及（N4）对支持工程师在CPS设计上运行不同分析的综合工具的需求。目标。我的研究项目将通过定义新颖的软件工程解决方案来支持工程师开发安全的CPS。为了实现这一长期目标，该计划考虑了四个短期目标，每个目标都针对一个研究挑战：（O1）开发识别CPS设计缺陷的方法，（O2）开发解释缺陷原因的方法，（O3）开发将人类工件转换为机器可处理格式的方法，以及（O4）开发集成所提出的解决方案的整体方法。通过开发新的基于伪造的测试技术、调试技术、基于模式的语言和对CPS设计进行不同分析的综合工具，将分别实现目标O1、O2、O3和O4。拟议研究的影响。这项研究将为加拿大提供显著的社会经济效益：提出的技术将使生产更安全、更可靠的CPS具有成本效益。它将巩固加拿大作为全球CPS设计领导者之一的地位，并在加拿大大学和CPS行业之间建立协同效应。

Background. Our society relies on cyber-physical systems (CPS). CPS are software-intensive systems that extensively interact with their physical environment. Most of our industries, including automotive, energy, and healthcare, rely on CPS for their daily operations. For example, CPS monitor and control our cars, power plants, and medical devices. According to the 2020 Industry Research CPS Market Report, in 2019 the global CPS market revenue was USD 52.4 billion and will reach USD 89.6 billion in 2025. Despite recent investments and improvements, CPS design is still complex, error-prone and, therefore, expensive. For example, the U.S. Defense Department estimates that the cost of the Block4 upgrade of the F-35 combat aircraft, which aims to improve the capabilities of the aircraft to counter emerging threats, reached USD 14.4 billion in 2020. In addition, CPS failures can also be catastrophic. For example, between 2018 and 2019, a defect in the Boeing 737 MAX resulted in the deaths of 346 people. Research Challenges. Engineers need automated support for CPS design. The research community has invested significant effort in developing automated techniques that support CPS design. Despite many successes, there are still many challenges that prevent extensive usage of these techniques in practice, especially for industrial models. This research program considers four research challenges: (N1) the need for automated techniques that search and detect flaws in the CPS design, (N2) the need for help in understanding the causes of the problems, (N3) the need for procedures that automatically translate human artifacts, such as CPS requirements, into machine-processable inputs, and (N4) the need for comprehensive tools that support engineers in running different analysis on the CPS design. Objectives. My research program will support engineers in developing safe CPS by defining novel software engineering solutions. To pursue this long term goal, this program considers four short term objectives, each targeting one of the research challenges: (O1) develop approaches to identify flaws in the CPS design, (O2) develop approaches to explain the causes of flaws, (O3) develop approaches to translate human artifacts into a machine-processable format, and (O4) develop holistic approaches to integrate the proposed solutions. Objectives O1, O2, O3, and O4 will be respectively achieved by developing novel falsification-based testing techniques, debugging techniques, pattern-based languages, and comprehensive tools running different analysis on the CPS design. Impact of Proposed Research. This research will provide remarkable socio-economical benefits to Canada: The proposed techniques will enable the cost-effective production of safer and more reliable CPS. It will consolidate Canada as one of the worldwide leaders in CPS design and build synergies between Canadian Universities and CPS industries.