CRII: CPS: Secure-by-Design Synthesis of Cyber-Physical Systems

包括无人驾驶汽车、联网医疗设备和智能制造在内的网络物理系统有望自主执行复杂任务，同时满足对安全和性能的严格要求。设计这种系统的一种新兴方法是自动生成捕获系统的期望功能的控制器。然而，系统执行控制的能力可能会受到恶意对手发起的网络和物理攻击的影响。该项目的目标是开发一个框架，用于自动生成控制行动，共同考虑安全，性能和安全。拟议的努力包含两个研究重点。第一个推力包括集成安全威胁的逻辑规范的网络物理系统（CPS）的方法，并验证CPS是否满足特定的对手模型下的规范。通过我们开发的一类新的行为游戏进行验证。第二个推力包括安全控制器的自动合成方法，满足在第一个推力中开发的规范。该项目旨在制定基于被动性的规则，用于组成多个控制律，以确保对多阶段攻击的安全性，以及根据当前对手行动的估计切换控制器的方法。这一努力将有助于开发在网络攻击存在下安全运行的CPS，如自动驾驶汽车和智能家居。该项目中开发的算法将在移动的平台上实现，以证明我们的方法的实际可行性，并确保成功和及时地过渡到实践。研究将由研究生研究助理进行，作为他们培训的一部分，而实施将与伍斯特理工学院的本科研究项目一起进行。研究结果将通过同行评审的出版物、知识产权披露和CPS虚拟组织（CPS-VO）发布的开源软件进行传播。

Cyber-physical systems, including unmanned vehicles, networked medical devices, and smart manufacturing, are expected to autonomously perform complex tasks, while satisfying strict requirements on safety and performance. An emerging approach to designing such systems is to automatically generate a controller that captures the desired functionality of the system. The ability of the system to execute the control, however, may be impacted by malicious adversaries mounting cyber and physical attacks. The goal of this project is to develop a framework for automatically generating control actions that jointly consider safety, performance, and security. The proposed effort contains two research thrusts. The first thrust consists of methodologies for integrating security threats into logical specifications for Cyber-physical systems (CPS), and verifying whether a CPS satisfies the specification under a given adversary model. Verification is performed via a new class of behavioral games that we develop. The second thrust consists of approaches for automatic synthesis of secure controllers that satisfy the specifications developed in the first thrust. The project aims to formulate passivity-based rules for composing multiple control laws to ensure security against multi-stage attacks, as well as methods for switching controllers based on the current estimate of the adversary actions.This effort will contribute to development of CPS, such as autonomous vehicles and smart homes, that operate safely in the presence of cyber attacks. Algorithms developed in the project will be implemented on mobile platforms in order to demonstrate the practical feasibility of our approach and ensure successful and timely transition to practice. Research will be conducted by graduate student research assistants as part of their training, while implementation will be done in conjunction with undergraduate research projects at Worcester Polytechnic Institute. Results will be disseminated through peer-reviewed publications, intellectual property disclosures, and open-source software release via CPS Virtual Organization (CPS-VO).