CPS: Medium: Collaborative Research: Against Coordinated Cyber and Physical Attacks: Unified Theory and Technologies

CPS：媒介：协作研究：对抗协调的网络和物理攻击：统一理论和技术

• 批准号: 1739886
• 负责人: Xiaofeng Wang
• 金额: $ 30万
• 依托单位: University of South Carolina at Columbia
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2017
• 资助国家: 美国
• 起止时间: 2017-09-01 至 2021-08-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1739886&HistoricalAwards=false
• 关键词: CPS; Medium; Collaborative; Research; Against

Coordinated cyber-physical attacks (CCPA) have been touted as a serious threat for several years, where "coordinated" means that attackers have complete knowledge of the physical plant and status, and sometimes can even create physical defects, to assist cyber attacks, and vice versa. In recent years, these attacks have crept from theory to reality, with attacks on vehicles, electrical grids, and industrial plants, which have the potential to cause destruction and even death outside of the digital world. CCPA raise a unique challenge with respect to cyber-physical systems (CPS) safety. Historically, technologies to defend cyber attacks and physical attacks are developed separately under different assumptions and models. For instance, cyber security technologies often require the complete profile of the physical dynamics and the observation of the system state, which may not be available when physical defects exist. Similarly, existing system control techniques may efficiently compensate for the physical damage, but under the assumption that the control software and the sensor data are not compromised. There is a lack of unified approaches against CCPA. With this observation, this project focuses on the development of unified models with coherent set of assumptions, supported by integrated technologies, upon which CCPA can be defended much more effectively.To establish theoretical foundations and engineering principles for resilient CPS architectures, this project will investigate unified models and platforms that represent the scientific understanding of resilient CPS against CCPA. Engineering of CPS will be addressed through the development and integration of complexity-reduced software architectures, along with their design principles, which lead to verifiable and certifiable architectures with higher level of system resilience. Technology of CPS will be addressed through the design of new attack detection, isolation, and recovery tools as well as timing and control techniques to ensure appropriate responses to CCPA. The proposed inherently interdisciplinary research will ensure predictable performance for resilient CPS, by leveraging the disciplinary advances in (i) the design and evaluation of robust fault-tolerant control systems yielding significantly enhanced levels of safety in highly unpredictable environments; (ii) the design and implementation of complexity reduction architecture yielding a significant reduction in the verification time from hours to seconds; (iii) the development of multi-rate sampled-data control and robust reachability-based attack detection techniques ensuring that the sensor data is reliable; and (iv) the development of cyber-physical co-adaptation that optimizes control performance and computation task scheduling to guarantee system safety and efficient recovery from CCPA. The target application of this project is unmanned aerial vehicles (UAVs). The research results will be evaluated in three different testbeds: UAV testbed, generic transportation model (GTM) aircraft, and power system virtual testbed (VTB). The technological advancement from this project will provide solutions for the safety and reliability issues faced by today's CPS and deliver dependable CPS that are applicable without sacrificing functionality or accessibility in complex and potentially hostile networked environment. The results of this project will be communicated in archival journal publications, conference venues and various workshops and lectures, and will be integrated at different academic levels.

几年来，协调的网络物理攻击（CCPA）已被吹捧为严重威胁，“协调”意味着攻击者完全了解物理植物和地位，有时甚至可以造成身体缺陷，以帮助网络攻击，反之亦然。 近年来，这些攻击从理论到现实都逐渐蔓延，对车辆，电网和工厂的攻击有可能在数字世界之外造成破坏甚至死亡。 CCPA在网络物理系统（CPS）安全方面提出了一个独特的挑战。 从历史上看，捍卫网络攻击和身体攻击的技术是在不同的假设和模型下分别开发的。 例如，网络安全技术通常需要对物理动力学的完整概况和系统状态的观察，当存在物理缺陷时可能无法使用。同样，现有的系统控制技术可能有效地补偿了物理损害，但是在控制软件和传感器数据不妥协的假设下。 缺乏针对CCPA的统一方法。 通过这一观察，该项目着重于具有一组连贯假设的统一模型的开发，并得到集成技术的支持，可以更有效地捍卫CCPA。为建立弹性CPS体系结构的理论基础和工程原理，该项目将调查对CCPA弹性CPA的科学了解的统一模型和平台。 CP的工程将通过开发和集成复杂性的软件体系结构及其设计原则来解决，这会导致具有更高水平的系统弹性水平的可验证和可认证的体系结构。 CP的技术将通过设计新的攻击检测，隔离和恢复工具以及时间和控制技术来解决，以确保对CCPA的适当响应。 拟议的固有跨学科研究将通过利用（i）（i）在高度不可预测的环境下的良好耐受耐受性控制系统的设计和评估来确保弹性CP的可预测性能； （ii）降低复杂性体系结构的设计和实施，验证时间从小时数大幅减少； （iii）开发多速率采样数据控制和基于可及性的攻击检测技术，以确保传感器数据可靠； （iv）开发网络物理共同适应，以优化控制性能和计算任务计划，以确保系统安全并有效从CCPA恢复。 该项目的目标应用是无人机（UAVS）。 研究结果将在三个不同的测试床上进行评估：无人机测试床，通用运输模型（GTM）飞机和电力系统虚拟测试床（VTB）。 该项目的技术进步将为当今CPS所面临的安全和可靠性问题提供解决方案，并提供可靠的CP，这些CP不适用，而无需牺牲复杂且潜在的敌对网络环境中的功能或可访问性。该项目的结果将在档案期刊出版物，会议场地以及各种讲习班和讲座中进行交流，并将在不同的学术层面进行整合。

项目成果

Asynchronous Lebesgue Approximation Model for Distributed Continuous-Time Nonlinear Systems

• DOI: 10.1109/cdc.2018.8619331
• 发表时间: 2018-12
• 期刊: 2018 IEEE Conference on Decision and Control (CDC)
• 作者: Ying Shen;Xiaofeng Wang;Zhengguang Wu

Lebesgue-Approximation-Based Model Predictive Control for Nonlinear Sampled-Data Systems with Measurement Noises

具有测量噪声的非线性采样数据系统的基于勒贝格近似的模型预测控制

• DOI: 10.23919/acc.2018.8431194
• 发表时间: 2018
• 期刊: American Control Conference
• 作者: Yang, Lixing;Wang, Xiaofeng
• 通讯作者: Wang, Xiaofeng

Optimizing Filter Size in Convolutional Neural Networks for Facial Action Unit Recognition

• DOI: 10.1109/cvpr.2018.00532
• 发表时间: 2018-01-01
• 期刊: 2018 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR)
• 作者: Han, Shizhong;Meng, Zibo;Tong, Yan
• 通讯作者: Tong, Yan