CRII: NeTS: Denoising Internet Delay Measurements using Weak Supervision

了解互联网的特征是互联网测量研究人员以及服务于全球数十亿用户的服务提供商和内容交付网络的关键目标之一。为此，已经开发了无数的测量工具和技术。尽管做出了这些努力，但由于测量噪声的存在，严重缺乏一个系统的框架来解释结果。本研究提出了一个新的框架来解决这一问题。该项目的主要目标是设计、开发和严格评估一个去噪延迟测量的框架。目标是通过两个关键的研究重点使去噪过程简单，自动和快速。在第一个推力中，目标是设计和开发框架，以利用机器学习的最新进展自动生成测量噪声标签（即地面真值数据）。第二个推力将扩展框架的能力，以自动化和快速的方式去除和修复噪声测量。将通过将框架应用于社区数据集，在实验室环境和现实环境中评估该框架的有效性。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。

Understanding the characteristics of the Internet is one of the key goals of Internet measurement researchers, and the service providers and content delivery networks that serve billions of users worldwide. To this end, a myriad of measurement tools and techniques have been developed. Despite these efforts, what is critically lacking is a systematic framework to interpret the results due to the presence of measurement noise. This research proposes to develop a new framework to solve this problem. The main goal of this project is to design, develop and rigorously evaluate a framework for denoising latency measurements. The objectives are to make the denoising process easy, automatic, and rapid via two key research thrusts. In the first thrust, the goal is to design and develop the framework to generate measurement noise labels (i.e., ground truth data) automatically leveraging recent advancements in machine learning. The second thrust will expand the capabilities of the framework to remove and repair the noisy measurements in an automated and rapid fashion. The efficacy of the framework will be evaluated in lab-based settings and in a real-world setting by applying it on community datasets.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.