Methodologies and Architectures for Probabilistic Real-Time Systems

受事件（例如警报、命令等）严格操作期限限制的计算系统我们称之为实时系统。在许多应用中，控制系统的响应时间可能对人员或基础设施（例如飞机，汽车或卫星）的安全产生关键影响。随着关键应用程序变得越来越复杂，实时系统的性能必须提高，以保证适当的安全标准。高性能实时系统的主要障碍是现代计算机体系结构的不可预测的时序行为。目前，为实时系统设计的软件需要保证在所需的时间范围内执行。根据底层硬件架构的复杂性，这种确定性可能很难或不可能达到。在此背景下，该项目将引入一种概率方法：通过在处理器和存储器架构中实现真正的随机行为，可以定义系统时序行为的概率度量。这将使实时软件工程与其他工程领域保持一致，这些领域使用组件故障概率执行风险分析。在这个项目中，我们将定义新的架构和设计方法，真正的概率实时系统，专注于航空航天应用。这种系统的成功实现将对关键系统的设计方式产生巨大影响，并且在实时软件的集成、验证和认证成本方面的潜在好处是巨大的。

Computing systems that are subject to strict operational deadlines from event (e.g. alarms, commands, etc.) to response are called real-time systems. These are employed in many applications where the response time of a control system can have a critical impact on the safety of people or infrastructure (e.g. airplanes, automobiles, or satellites). With critical applications become more and more complex, the performance of real-time systems has to increase to guarantee appropriate safety standards. The main obstacle to high-performance real-time systems is the unpredictable timing behaviour of modern computer architectures. Currently, software designed for real-time systems needs to be guaranteed to be executed within the required time frame. Depending on the complexity of the underlying hardware architecture, this level of certainty might be difficult or impossible to reach. Within this context, this project will introduce a probabilistic approach: by enabling true randomized behaviour in processors and memory architectures, one can define probabilistic metrics for the timing behaviour of a system. This would bring real-time software engineering in line with other engineering domains, that perform risk analysis using component failure probabilities. In this project we will define novel architectures and design methodologies for truly probabilistic real-time systems, focusing on aerospace applications. Successful implementation of such systems will have tremendous impact on the way critical systems are designed, and the potential benefits in terms of cost of integration, verification, and certification of real-time software are enormous.