Note: This blog was contributed by REU student Zack Murry, an undergraduate in Computer Science at the University of Missouri. We were fortunate to have Zack working with us alongside three other REU students for the Summer 2024 quarter. Zack has submitted his work to the SC 24 Poster Competition this coming Fall. Good luck, Zack!

The advent of new technologies such as machine learning, the Internet of Things (IoT), and high-quality video streaming presents increasing challenges to the Internet’s infrastructure, as they demand massive amounts of data to be produced, uploaded, and processed. One bottleneck in this process is the “last mile” of the Internet connection, or the link between user devices and the broader high-speed Internet infrastructure.

Fifth-generation (5G) cellular networks promise to improve last-mile connectivity by increasing upload and download speeds, providing more consistent connections, and allowing more devices to be connected simultaneously. 5G can achieve speeds of up to 10 gigabits per second, which is 10 times faster than 4G networks [1]. This is primarily due to the use of higher-frequency radio waves and more sophisticated encoding techniques.

In collaboration with the Agricultural and Rural Wireless Living Lab (ARA), we developed infrastructure to measure the performance of their 5G networks and to test new 5G-enabled computation paradigms. ARA is an at-scale platform for advanced wireless research deployed across the Iowa State University campus, the city of Ames, Iowa, and surrounding research and producer farms [2].

We have integrated six single-board computers into ARA’s 5G wireless platform. These devices collect real-time network speed data in two ways, measuring speeds both to distant servers and to other devices on the 5G network. The former models the overall speeds users could expect, whereas the latter observes only the last-mile performance. Both of these metrics are important because they can provide practical insight into day-to-day speeds and allow for comparisons with the existing commercial infrastructure in the area.

FLOTO enables this work by providing a platform for remotely managing these devices, executing jobs, and collecting data. Instead of being physically present to set up, debug, and run experiments, we use FLOTO to conduct research from another state. This means we can use the one-of-a-kind ARA testbed to make discoveries without intricate coordination with their on-site team. Additionally, by supporting containerized experiments, FLOTO ensures that our work can be easily reproduced, as each test can be launched on any device in a matter of minutes through the dashboard.

In addition to the above experiments, we aim to compare different 5G technologies and wavelengths for common use cases and variables, including their resilience to adverse weather conditions.

We’re also experimenting with new 5G-enabled distributed computing paradigms, bringing tools that have traditionally been used with wired connections between servers to small devices connected on 5G networks. Wireless networks have traditionally been too slow to support large data-processing tasks, but 5G presents an opportunity to explore new ways of multi-device computation.

With the results of this project, we hope to increase observability of the capabilities of 5G, particularly in rural and agricultural settings, so that consumers, companies, and governments can make better-informed decisions in designing and using 5G wireless networks.

---

[1] https://aws.amazon.com/what-is/5g/
[2] https://arawireless.org
[3] https://community.element14.com/learn/learning-center/the-tech-connection/w/documents/27878/understanding-the-5g-spectrum