CRII: SHF: Leveraging Synthesis for Dynamic Design Space Analysis

设计大型、复杂的系统，要求一定的功能和服务质量（QoS）的目标仍然是一个重要的工程问题。设计空间探索和权衡分析可以发挥关键作用，揭示人类设计师可能错过的设计，阐明合理和荒谬的权衡，并帮助决策者平衡设计决策对不同利益相关者的权衡。尽管其重要性，系统的权衡设计空间分析可能是复杂系统设计中最难以捉摸的活动之一，因为进行交易空间分析所需的成本，工作量和纪律可能非常高。 本研究设计了一套全自动化的技术，以促进系统的软件系统的设计权衡空间的分析。通过降低贸易空间对复杂软件设计进行分析的成本，并使其更具可扩展性，该项目使充满活力的软件行业能够提高其产品的质量。该研究项目包括分析和实证两个部分。分析方法涉及开发模型和算法，其可以通过自动合成来自设计者的高级输入的设计解决方案的空间来发现设计备选方案，通过自动使每个设计备选方案经受不同评估操作的执行来比较分析这样的设计空间，并且通过利用各种优化机制来促进对设计空间演化的支持。经验评估部分包含多个步骤进行实验：通过测试台进行的受控实验和在真实世界系统中进行的实验。该项目通过使合成驱动的软件设计更具可扩展性和成本效益来推进最先进的技术。从这项研究中产生的原则将有助于扩大我们的应用领域和设置，其中基于综合的方法来设计权衡空间分析可以实际使用的理解。

Designing large, complex systems that demand certain functional and quality of service (QoS) objectives remains a significant engineering problem. Design space exploration and tradeoff analysis can play a pivotal role by revealing designs that human designers might miss, illuminating sensible and nonsensical tradeoffs, and helping decision-makers to balance tradeoffs that design decisions impose on diverse stakeholders. Despite its importance, systematic tradeoff design space analysis is perhaps one of the most elusive activities in the design of complex systems, as the cost, effort, and discipline needed to conduct tradespace analysis can be extremely high. This research devises a suite of fully automated techniques to foster systematic analysis of design tradeoff spaces of software systems. By making tradespace analysis of complex software designs less expensive to conduct and more scalable, this project enables the vibrant software industry to improve the quality of its products.The research project contains both analytical and empirical components. The analytical approaches involve developing models and algorithms that can discover design alternatives via automatically synthesizing spaces of design solutions from designers' high-level inputs, comparatively analyze such design spaces via automatically subjecting each design alternative to the execution of diverse evaluation operations, and facilitate support for design space evolution through leveraging a variety of optimization mechanisms. The empirical evaluation component contains conducting experiments in multiple steps: controlled experiments by means of test beds and experiments in the context of real-world systems. The project advances the state-of-the-art by making synthesis-driven software design more scalable and cost effective. The principles resulting from this research will help expand our understanding of the application domains and settings in which synthesis-based approach to design tradeoff space analysis can be practically used.