GRK 1765: System Correctness under Adverse Conditions - SCARE

研究培训小组SCARE研究的目标是将计算机系统置于与其合作的环境中，即感知、控制和配备前所未有的功能。系统正确性是指环境和系统之间的配合满足了期望的行为属性。这种关系依赖于对环境和系统组成部分的某些假设。研究培训小组系统地研究在不利的、只有部分可预测的条件下的系统正确性问题，这些条件会影响系统的行为、系统上下文和为验证正确性而做出的假设。SCARE将考虑三个方面的不利条件，包括单独的和综合的：A.有限的知识。不可预知的behaviour.C。环境和系统结构的变化。这三个方面的不利条件是考虑以下研究主题：1。形式化建模技术。验证和分析技术。建设性的技术。研究培训学校SCARE的主要目的是研究系统正确性的概念，以保证在这种不利条件下系统行为的鲁棒性。在第二阶段，SCARE将继续进行这一总体研究议程，但将从两个方面扩展验证和构建技术的范围，这两个方面对最近的网络物理应用至关重要：处理复杂和可能不规则的搜索空间的解决方案，以及由于安全攻击可能导致的功能正确性损失。为此，我们还追求以下目标：*嵌入式人工智能，特别是机器学习。*应用程序内的安全分析。科研培训学校的督导以“拉平”、“纽带”、“进步”、“交流”四个要素为基础。这些因素支持博士成为独立的研究人员，能够在自己的动机驱动下研究复杂的问题，并自主地与国际研究界互动。

The Research Training Group SCARE addresses computerised systems that are placed in an environment with which they cooperate, i.e., sense, control, and equip with unprecedented functionality. System correctness means that the cooperation between environment and system satisfies desired behavioural properties. This relationship depends on certain assumptions about the environment and the components of the system. The research training group systematically investigates the problem of system correctness under adverse, only partially predictable conditions which can influence the behaviour of the system, the system context, and the assumptions made for verifying correctness. SCARE will consider three aspects of adverse conditions, both individually and in their combination: A. Limited knowledge.B. Unpredictable behaviour.C. Changing structure of environment and system.These three aspects of adverse conditions are considered under the following research themes:1. Formal modeling techniques.2. Verification and analysis techniques.3. Constructive techniques.The main aim of the research training school SCARE is research into notions of system correctness that guarantee robustness of the system behaviour under such adverse conditions. In its second phase, SCARE will continue to pursue this general research agenda, but will extend the scope of both verification and construction techniques by two aspects crucial to recent cyber-physical applications: the handling of complex and possibly irregular search spaces for solutions and the possible loss of functional correctness due to security attacks. To this end, we additionally pursue the following: * Embedded Artificial Intelligence, in particular machine learning.* Security analysis within applications.Supervision in the research training school is based on the four elements of "Levelling", "Bonding", "Progress", and "Exchange". These elements support the doctorands in becoming independent researchers able to investigate complex problems driven by their own motivation and to interact autonomously with the international research community.