Resilience meets secure networked control

网络化控制系统是许多技术的核心组成部分，通常构成关键基础设施的骨干。许多应用程序还涉及敏感和机密数据的处理，如商业秘密或个人数据。年轻但新兴的加密控制领域通过将控制方案与密码系统相结合来解决这个问题，从而在整个控制回路中保证机密性。然而，虽然现代密码学提供了强大的弹性对攻击的数据的机密性，现有的实现加密控制的性能不足的实际使用或基于非正式的特设的方法，不反映在现代密码学和IT安全研究的最新技术。因此，年轻的技术具有很大的实际意义，现在已经达到了一个阶段，控制工程师和密码学家之间的跨学科合作是必要的，以便能够构建系统的性能和安全性大大超出了最先进的国家。在这个项目中，我们将开发专用的加密技术，是针对现代网络控制系统的要求，我们将相应地采用相应的控制算法，并使它们尽可能与现代密码技术提供的功能兼容。为此，我们将研究几个具体的想法设计的专用近似同态加密方案与后量子安全，算法修改各种现代控制算法，涉及机器学习和优化，以及它们的可实现性与现代密码安全计算技术。

Networked control systems are central building blocks for many technologies and often form the backbone of critical infrastructure. Many applications further involve the processing of sensitive and confidential data such as business secrets or personal data. The young but emerging field of encrypted control addresses this issue by combining control schemes with cryptosystems in such a way that confidentiality is guaranteed during the entire control loop. However, while modern cryptography offers strong resilience against attacks on the confidentiality of data, existing realizations of encrypted control are either insufficiently performant for practical use or are based on an informal ad hoc approach that does not reflect the state of the art in modern cryptography and IT security research. Thus, the young technology with great practical relevance has now reached a stage where interdisciplinary collaboration between control engineers and cryptographers is necessary in order to being able to construct systems whose performance and security goes significantly beyond the state of the art. In this project, we will develop special-purpose cryptographic techniques that are tailored to the requirements of modern networked control systems, and we will accordingly adopt the corresponding control algorithms and make them as compatible as possible with the functionalities provided by modern cryptographic techniques. To this end, we will investigate several concrete ideas on the design of special-purpose approximate homomorphic encryption schemes with post quantum security, algorithmic modifications to various modern control algorithms involving machine learning and optimization, and their realizability with modern cryptographic secure computation techniques.