CSR --- EHS: The Design and Self-Organization of Component-based Systems for Dependable and Predictable Embedded Computing Environments

这项研究解决了几个关键的不足，现有的系统中使用的实时和嵌入式计算。这些缺陷包括：（1）僵化的系统结构，导致系统服务的组织方式和彼此隔离的可扩展性，（2）应用程序的需求和系统提供的服务之间的语义差距，以及（3）缺乏对将时间作为一流资源进行管理的支持。 该研究通过开发一个基于组件的系统，在一系列硬件平台上安全，可预测和高效地执行特定于应用程序的服务，解决了这些缺陷。在安全方面的研究旨在确保可靠性的软件组件的集合，使系统的完整性不受损害，并在何种程度上错误或不可信的代码可能会影响其他组件的资源使用是有限的。该研究正在开发一个系统，是灵活的隔离边界周围的组件服务的位置。 使用硬件隔离功能（例如，分页、分段、虚拟化技术）。权衡组件之间的通信开销进行了研究，沿着的机制和政策，动态调整组件服务的隔离粒度（因此组件间的通信成本），以确保可预测的服务延迟。这项工作预计将成为正在进行的系统设计研究的基础，以确保高度的软件可靠性，至少包括故障隔离，同时保证其及时执行。预计这将为在广泛的嵌入式计算领域中实现未来系统奠定基础，特别是那些包含网络物理系统的系统。

This research addresses several key deficiencies of existing systems used in real-time and embedded computing. These deficiencies include: (1) a rigid system structure, leading to inflexibility in the way system services are organized and isolated from each other, (2) a semantic gap between the needs of applications and the services offered by the system, and (3) a lack of support for managing time as a first-class resource. The research addresses these deficiencies by developing a component-based system for safe, predictable and efficient execution of application-specific services on a range of hardware platforms. In terms of safety the research seeks to ensure the dependability of a collection of software components, so that system integrity is not compromised, and the extent to which errant or untrustworthy code may impact the resource usage of other components is limited.The research is developing a system that is flexible in its placement of isolation boundaries around component services. The uses of hardware isolation features (e.g., paging, segmentation, virtualization technologies) on processors such as the x86 and ARM are investigated. Trade-offs in the communication overheads between components are studied, along with mechanisms and policies that dynamically adjust the isolation granularity of component services (hence inter-component communication costs) to ensure predictable service delays. This work is expected to form the basis for ongoing research into the design of systems that ensure a high degree of software dependability, including fault isolation at a minimum, while guaranteeing their timely execution. It is expected this will lay foundations for implementing future systems in a wide range of embedded computing domains, particularly those that encompass cyber-physical systems.