Given the widespread deployment of cyber-physical systems and their safety-critical nature, reliability and security guarantees offered by such systems are of paramount importance. While the security of such systems against sensor attacks have garnered significant attention from researchers in recent times, improving the reliability of a control software implementation against transient environmental disturbances need to be investigated further. Scalable formal methods for verification of actual control performance guarantee offered by software implementations of control laws in the face of sensory faults have been explored in recent work [20]. However, the formal verification of the improvement of system reliability by incorporating sensor fault mitigation techniques like Kalman filtering [29] and sensor fusion [18, 52] remains to be explored. Moreover, system designers face complex tradeoff choices for deciding upon the usage of fault and attack mitigation techniques and scheduling them on available system resources as they incur extra computation load. In the present work, our contributions are threefold. We formally analyze the actual performance guarantee of control software implementations enabled with additional fault mitigation techniques. We consider task-level models of such implementations enabled with security and fault tolerance primitives and construct a timed automata-based model which checks for schedulability on heterogeneous multi-core platforms. We leverage these methodologies in the context of a novel Design-Space-Exploration (DSE) framework that considers target reliability and security guarantees for a control system and computes schedulable design options while considering well-known platform-level security improvement and fault mitigation techniques. We validate our contributions over several case studies from the automotive domain.

中文翻译：

---

网络物理系统的可靠和安全的设计-空间-探索

鉴于网络物理系统的广泛部署及其对安全至关重要的性质，此类系统提供的可靠性和安全保障至关重要。尽管此类系统针对传感器攻击的安全性近年来引起了研究人员的极大关注，但需要进一步研究提高控制软件实现对瞬态环境干扰的可靠性。在最近的工作中，已经探索了可扩展的形式化方法，用于验证在面对感官故障时控制律的软件实现所提供的实际控制性能保证[20]。然而，通过结合传感器故障缓解技术（如卡尔曼滤波 [29] 和传感器融合 [18, 52]）对系统可靠性改进的形式验证仍有待探索。此外，系统设计人员面临着复杂的权衡选择，以决定故障和攻击缓解技术的使用以及在可用系统资源上调度它们，因为它们会产生额外的计算负载。在目前的工作中，我们的贡献是三倍的。我们正式分析了使用附加故障缓解技术启用的控制软件实现的实际性能保证。我们考虑启用了安全性和容错原语的此类实现的任务级模型，并构建了一个基于定时自动机的模型，用于检查异构多核平台上的可调度性。我们在新颖的设计-空间-探索 (DSE) 框架的背景下利用这些方法，该框架考虑控制系统的目标可靠性和安全保证，并计算可调度的设计选项，同时考虑众所周知的平台级安全改进和故障缓解技术。我们在汽车领域的几个案例研究中验证了我们的贡献。