IMR: MM-1A: ADDRESS: Augment, Denoise and Debias cRowdsourced mEasurements for Statistical Synthesis of internet access characterization

美国遭受着数字不平等的困扰，这种不平等在多个维度上都有延伸。几个人口和社会经济因素，如种族、民族、收入以及人口密度，往往是互联网可用性和质量的指标。为了充分弥合互联网接入鸿沟并以最佳方式分配可用资金，政策制定者需要在细粒度的地理粒度上了解互联网接入和质量的状况。该项目的目标是充分利用众包的互联网测量数据，以尽可能高的精度估计固定互联网质量在不同地理空间粒度级别上的分布。为此，该项目计划做出以下贡献：(I)开发一个宽带产品工具，将这些数据与众包测量和社会经济数据联系起来，以预测用户订阅计划；(Ii)开发能够检测噪声数据点的新方法和算法，例如报告WiFi接入或远程对等链路中的瓶颈的速度测试测量，这会影响对互联网质量的理解，(Iii)提出新的统计技术，使用产生的数据集输出去偏的数据集，消除众包互联网测量数据中固有的空间、时间和人口偏见。数字不平等在美国继续存在，新冠肺炎大流行导致的在家工作和远程教育加剧了数字不平等。这项工作的成果将用于报告一个区域的互联网质量分布情况，并预测数据较少的区域的质量分布情况。这项工作还将揭示对美国数字鸿沟影响最大的人口变量。这个项目的成果可以用来告知地方、州和联邦政府必须在哪里进行投资，以确保所有美国人都能接触到高质量的互联网。将建立一个项目网站，其中将包含有关研究方法、模型和结果的信息。可能的URL是Broadband.cs.ucsb.edu。该网站将在项目期间和至少一年后维护。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。

The U.S. suffers from digital inequality that stretches in multiple dimensions. Several demographic and socioeconomic factors, such as race, ethnicity, and income, as well as population density, are often indicators of Internet availability and quality. To fully bridge the Internet access divide and best allocate available funding, policymakers need to understand the state of Internet accessibility and quality at fine-grained geographical granularity. The goal of this project is to make the best use of crowdsourced Internet measurement data to estimate the distribution of fixed Internet quality at different levels of geospatial granularity with as high accuracy as possible.To do so, the projects plans to make the following contributions: (i) to develop a Broadband Offerings Tool that aggregates ISP broadband plan offerings and links that data with crowdsourced measurements and socioeconomic data to predict user subscription plans, (ii) to develop new methodologies and algorithms that can detect noisy data points, such as speed test measurements that report bottlenecks in WiFi access or remote peering links, which bias the understanding of Internet quality, (iii) to propose new statistical techniques that use the resulting dataset to output a debiased dataset that removes the spatial, temporal, and demographic biases inherent in crowdsourced Internet measurement data.Digital inequality continues to persist in the U.S., deepened by the work-from-home and remote schooling resulting from the COVID-19 pandemic. The outputs of this work will be used to report the distribution of Internet quality in a region and predict that of regions for which less data is available. This work will also reveal demographic variables that most influence the U.S. digital divide. The outcomes of this project can be used to inform local, state and federal governments about where investments must be made to ensure all Americans have access to high quality Internet.A project website will be established, which will contain information about research methodology, models and outcomes. The likely URL is broadband.cs.ucsb.edu. The website will be maintained for the duration of the project, and at least a year thereafter.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.