Fault-Tolerant Energy Management for Highly Dependable Real-Time Embedded Systems

实时嵌入式系统（RTES）在世界范围内的日常生活中无处不在和普遍存在，目前在从智能手机到智能汽车和航空/航天器的众多应用中发现。随着使用RTES的设备数量的增加，以及对更高性能的需求，需要改进的能源管理解决方案，以在不牺牲性能的情况下抑制能耗。另一方面，容错也是普适和普适计算平台中的主要关注点。该项目将为RTES开发新的能量管理算法，同时满足容错要求。这些发现将从根本上改变RTES的设计方式，使其能够满足“绿色和高度可靠”计算平台的性能需求。因此，这项工作的结果不仅对该行业的财务收益，而且对环境的健康也会产生巨大的全球影响。此外，该项目通过为本科生和研究生研究人员提供实践独特的研究经验，使他们能够应用理论学习，并激励他们探索计算机科学的职业生涯，将研究和教育结合起来。最后，这项工作旨在奠定基础，为未来的技术进步，可以增加这个国家的竞争优势，同时也提供社会效益。在这个项目中，一个新的调度框架将被设计为提高高可靠性的实时嵌入式系统（RTES）的性能下的容错要求，利用现代电源管理功能，可在异构多核处理器。本论文的主要目标是：（1）针对实时嵌入式系统（RTES）的容错需求，提出一种新的能量管理方案，以降低系统的能耗;（2）在严格的能量预算约束下，提出一种调度框架，以最大限度地提高容错RTES的可行性;（3）将该调度框架扩展到众核平台。该项目的新奇之处在于它结合了RTES设计的三个关键维度，即，能源效率，容错和服务质量（QoS）到一个单一的，统一的框架，以减少能源消耗，以及最大限度地提高可行性，同时满足从系统层面的RTES的QoS。这个奖项反映了NSF的法定使命，并已被认为是值得通过使用基金会的智力价值和更广泛的影响审查标准进行评估的支持。

Real-time embedded systems (RTES) are ubiquitous and pervasive in everyday life worldwide, currently found in numerous applications ranging from smart phones to intelligent cars and air/spacecraft. The increased number of devices that use RTES, in conjunction with the demand for ever greater performance, call for improved energy management solutions to stem energy consumption without sacrificing performance. On the other hand, fault tolerance has also been a major concern in pervasive and ubiquitous computing platforms. The proposed project will develop new energy management algorithms for RTES while satisfying the fault tolerance requirement. The findings will radically shift how RTES are designed to be able to meet performance demands on a “green and highly dependable” computing platform. As such, the results of this work can also have an enormous global impact not only on the industry’s financial gains, but also on the environment’s health. Furthermore, this project integrates research and education by providing undergraduate and graduate researchers with hands-on unique research experiences where they can apply theoretical learning and can inspire them to explore a career in computer science. Finally, this work intends to lay the groundwork for future technological advances which can increase the competitive edge of this country while also providing societal benefits.In this project, a novel scheduling framework will be designed to improve the performance of highly dependable real-time embedded systems (RTES) under fault tolerance requirements, exploiting modern power-management features available in heterogeneous multi-core processors. Three Specific Aims will be addressed: (1) to develop novel energy management schemes that can reduce energy consumption for real-time embedded systems (RTES) under fault tolerance requirement; (2) to develop algorithms that can create a scheduling framework targeting maximizing the feasibility of fault tolerant RTES under hard energy budget constraint; and (3) to extend this scheduling framework to many-core platforms. The novelty of the project lies in the fact that it combines the three critical dimensions of RTES design, i.e., energy efficiency, fault tolerance, and Quality of Service (QoS) into a single, unified framework to reduce energy consumption as well as to maximize the feasibility while satisfying the QoS for RTES from the system level.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.