CSR-EHS: An Extended Theory for Temporal Composition of Distributed Real-Time Computing Systems

在下一个十年中，计算领域的一个基本进步将出现在逻辑和物理领域之间的接口上。未来的计算系统将看到计算、通信和传感的日益融合。它们将以大规模分布为特征，并将增强物理环境，在实时和空间的限制下与之互动。这个项目发展了一个计算这种未来系统实时能力的基本理论，它量化了它们及时工作的能力。在给定应用程序流程图、时间约束和资源拓扑的情况下，建立了组成实时容量表达式的数学基础。这个基础利用一小组组合规则来获得应用程序感知的容量表达式。该项目还考虑了当系统组件变得无穷小而应用程序所需的组件数量增长到无穷大时的限制。一旦为运行一组应用程序的分布式系统构建了容量区域，就可以执行各种优化，例如在容量区域内找到最大吞吐量点。

One of the fundamental advances in computing in the next decade will occur at the interface between the logical and physical realms. Future computing systems will see increased convergence of computation, communication, and sensing. They will be marked by massive distribution and will augment the physical environment, interacting with it under constraints of real time and space. This project develops a fundamental theory for computing the real-time capacity of such future systems which quantifies their ability to do timely work. A mathematical foundation is developed for composing real-time capacity expressions given application flow graphs, time constraints and resource topology. This foundation utilizes a small set of composition rules to arrive at application-aware capacity expressions. The project also considers the limits when system components become infinitesimal and the number of components needed for an application grows to infinity. Once a capacity region is constructed for a distributed system running a set of applications, various optimizations can be performed such as finding the point of maximum throughput within the capacity region.