5G tracker: a crowdsourced platform to enable research using commercial 5g services
5G tracker: a crowdsourced platform to enable research using commercial 5g services
复制标题
5G 跟踪器：一个众包平台，可使用商业 5G 服务进行研究
DOI：
10.1145/3405837.3411394
复制
发表时间：
2020
期刊：
影响因子：
--
通讯作者：
Zhi
中科院分区：
文献类型：
--
作者：
A. Narayanan;Eman Ramadan;Jacob Quant;Peiqi Ji;Feng Qian;Zhi
虽然5G为研究提供了许多机会，但大多数研究都是在受限的实验环境下进行的，或者是由5G运营商私下进行的。即使在商用5G网络推出一年后，由于缺乏公开可用的工具和数据集，对商用5G的研究也受到限制。在本文中，我们提出了5G跟踪器——一个众包平台，旨在帮助研究人员收集和利用大规模5G数据集。该平台包括一个安卓应用程序，可以记录为5G网络和研究量身定制的被动和主动测量。我们已经使用5G Tracker超过8个月，在此期间，我们收集了超过400万个数据点，在美国的多个5G运营商中消耗了超过50tb的蜂窝数据。我们的经验表明，5G性能除了受到位置的影响外，还受到用户移动水平、方向、天气、位置动态（如移动车辆）以及环境特征（如柱子、树叶和建筑物）等多个用户侧环境因素的影响。这在一定程度上是因为毫米波信号（被认为是主流5G的关键）对障碍物和用户移动性高度敏感。这些观察结果强调了构建上下文感知的5G性能预测模型的必要性，该模型可以为抢占式切换、多路径调度、塔放置和“5G感知”应用程序开发等各层的决策提供指导。最后，我们通过构建一个基于我们收集的数据驱动的首创交互式5G覆盖地图应用程序作为案例研究来展示我们平台的实用性，该应用程序可在https://5gophers.umn.edu上公开获取。
While 5G has offered many opportunities for research, the majority of studies have been conducted with constrained experimental settings or done privately by 5G operators. Even a year after the launch of commercial 5G networks, research over commercial 5G has been limited due to the lack of publicly available tools and datasets. In this paper, we propose 5G Tracker - a crowdsourced platform intended to aid researchers in collecting and leveraging large-scale 5G datasets. This platform includes an Android app that records passive and active measurements tailored to 5G networks and research. We have been using 5G Tracker for over 8 months, during which time we have collected over 4 million data points, consuming over 50 TB of cellular data across multiple 5G carriers in the U.S. Our experience shows that 5G performance is affected by several user-side contextual factors besides location such as user mobility level, orientation, weather, location dynamics (e.g., moving vehicles), and environmental features such as pillars, foliage, and buildings. This is partly because mmWave signals (considered key to mainstream 5G) are known to be highly sensitive to obstructions and user mobility. These observations highlight the need to move towards building context-aware 5G performance prediction models that can provide guidance for decisions at various layers such as preemptive handoff, multi-path scheduling, tower placement, and "5G-aware" application development. Finally, we showcase the utility of our platform by building a first of kind, interactive 5G coverage mapping application as a case study driven by the data we collected, which is publicly available at: https://5gophers.umn.edu.