CAREER: Taming Wireless Devices Cross-Layer Errors with Assistive Networked Edges

智能手机、电脑和传感器等无线设备在我们的日常生活和现代社会中无处不在，在工作效率、医疗保健、自动化控制等方面创造了前所未有的好处。由于不完善的电子设备和嘈杂的环境，无线设备固有的缺陷可能导致计算、缓存和通信中的多方面数据错误（C3）。这些误差被普遍认为是有害的，所以目前的误差控制主要是为了消除绝对误差。然而，数据错误可能是良性的，甚至是有益的（例如，将错误引入梯度可能有助于机器学习模型逃避局部最优），而现有的反应性和非功能性研究未能将数据错误转化为有益的。因此，该项目的研究目标是主动地收集、呈现和控制无线设备C3中的数据错误，从而在能效、吞吐量、数据隐私等方面获得显著的性能提升。此外，研究工作将与教育创新相结合，通过开发新的实验室，讲座内容，外展演示和新颖的本科/研究生共同学习教学法。该项目的成功完成将增强该国无线工作人员的多样性，促进社区外展，并创造对未来无线应用（例如人工智能和智能健康）具有变革性的无线技术的根本性创新。拟议的研究活动包括跨应用层的硬件和软件设计的协同作用，无线设备中的嵌入式存储器和前端无线电以及无线网络边缘的网络协议。预期的结果包括：(1)在运行时收集并呈现错误到特定的数据位（即，应用程序感知的），以实现系统性能提升。(2)利用隐式控制信号将错误与上下文感知（如通道条件）整合和控制，以优化自适应。(3)开发一套边缘网络协议，方便错误数据的控制。与此同时，拟议的教育活动包括创建硬件和应用感知无线项目、实验室、讲座内容和扩展演示。这些材料将按照设计的本科生/研究生共同学习教学法和计划的本科生研究途径全面交付。还将设计和开展各种推广活动，以低收入和农村的K-12学生以及第一代大学生为目标，这是PI所在州独特的代表性不足的群体。预期结果包括：(1)通过提供更清晰的学科重点，为K-12学生未来的大学学习做好准备。(2)增加我国无线技术劳动力的多样性和人口。该项目由电气、通信和网络系统部、工程局和刺激竞争研究的既定计划（EPSCoR）共同资助。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

Wireless devices such as smart phones, computers and sensors are ubiquitous in our daily life and modern society, creating unprecedented benefits in work efficiency, healthcare, automation control and many more. Due to the imperfect electronics and noisy environments, wireless devices are inherently faculty which can result in multifaceted data errors in computing, caching, and communications (C3). These errors have been widely deemed harmful, so the state-of-the-art on error control mainly target for absolute error removal. Yet, data errors can be benign or even beneficial (e.g., introducing errors to gradients may help the machine learning models to escape local optima), while existing research that are reactive and non-functional fail to turn data errors into good. Hence, the research objective of this project is to proactively harvest, render, and control data errors across C3 of wireless devices for significant performance gains in energy efficiency, throughput, data privacy, etc. Moreover, the research efforts will be coupled with educational innovations through the development of new laboratories, lecture contents, outreach demos and a novel undergraduate/graduate co-learning pedagogy. The successful completion of this project will enhance diversity in the wireless workforce in the nation, promote community outreach and create fundamental innovations of wireless technologies that are transformative to future wireless applications (e.g., AI and smart health).The proposed research activities include a synergy of hardware and software designs across layers of application, embedded memory and front-end radio in wireless devices as well as network protocols on wireless network edges. Expected outcomes include: (1) Harvest and render errors to specific data bits (i.e., application-aware) at runtime to realize system performance gains. (2) Integrate and control errors with contextual awareness (e.g., channel condition) for optimized adaptation using implicit control signaling. (3) Develop a suite of edge networking protocols to facilitate control of erroneous data. In parallel efforts, the proposed education activities include the creation of hardware- and application-aware wireless projects, laboratories, lecture contents and outreach demos. These materials will be delivered holistically by following a designed undergraduate/graduate co-learning pedagogy and a planned undergraduate research pathway. Various outreach activities will be also designed and performed to target low-income and rural K-12 students and first-generation college students – unique under-represented groups at PI’s home state. Expected outcomes include: (1) Prepare K-12 students for future collegiate study by offering a clearer disciplinary focus. (2) Increase the diversity and the population of our nation’s workforce in wireless technologies.This project is jointly funded by the Division of Electrical, Communications and Cyber Systems, Directorate of Engineering, and the Established Program to Stimulate Competitive Research (EPSCoR).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.