CCF: Medium: Enabling Real-Time Quantitative Decision Making over Streaming Data

有效的物联网(IoT)系统的一个关键组件是实时做出决策的能力，以响应其接收的数据。虽然在不同的应用程序中制定决策的确切逻辑需要特定于领域的洞察，但它通常依赖于以高效和增量的方式计算大型数据流的量化摘要。由于数据量巨大以及对可用内存和响应时间的严格限制，对所需逻辑进行编程是具有挑战性的。该项目旨在帮助物联网程序员应对这一挑战，设计一种具有适合处理流数据的自然和高级结构的查询语言，由编译器和运行时系统支持，以便于部署，同时满足期望的准确性、内存占用和实时响应的限制。查询语言的设计借鉴了两种不同范例的见解：用于处理流数据的关系查询语言的扩展，以及用于运行时监控和同步编程的基于状态的语言。语言结构的新颖集成允许程序员给予输入数据流一个逻辑层次结构，并且还使用关系结构来按键划分输入数据并合并来自不同传感器的数据流。虽然近似和流算法的理论允许在编译过程中在语言特性、准确性和处理时间之间进行权衡，但诸如ApacheStorm这样的分布式平台用于高性能处理。实验评估使用了两个应用领域：生理患者数据的设备监控和异常的网络流量动态监控。关于可预测的实时决策的课程模块正在开发中，可用于数据科学的本科课程、数据库系统和网络物理系统的研究生课程、暑期学校和教程。该奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

A key component of an effective Internet of Things (IoT) system is the ability to make decisions in real-time in response to data it receives. While the exact logic for making decisions in different applications requires domain-specific insights, it typically relies on computing quantitative summaries of large data streams in an efficient and incremental manner. Programming the desired logic is challenging due to the enormous volume of data and hard constraints on available memory and response time. This project aims to assist IoT programmers in meeting this challenge by designing a query language with natural and high-level constructs suitable for processing streaming data, supported by a compiler and run-time system to facilitate deployment while meeting the constraints of desired accuracy, memory footprint, and real-time response.The design of the query language draws upon insights from two distinct paradigms: extensions of relational query languages for handling streaming data, and state-based languages for runtime monitoring and synchronous programming. The novel integration of linguistic constructs allows the programmer to impart input data stream a logical hierarchical structure and also employ relational constructs to partition the input data by keys and to merge data streams from different sensors. While theory of approximation and streaming algorithms allows exploration of trade-offs among language features, accuracy, and processing time during compilation, distributed platforms such as Apache Storm are used for high performance processing. Two application domains, device monitoring for physiological patient data and dynamic monitoring of network traffic for anomalies, are used for experimental evaluation. A course module on predictable real-time decision making is being developed, and can be used in an undergraduate course on data science, in graduate courses on database systems and cyber-physical systems, in summer schools, and as tutorials.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.