New Methods for Residential Broadband Monitoring

Accurate data regarding who actually has access to the Internet, and at what speeds, is a widely known obstacle to closing the digital divide. Existing techniques (FCC maps; Ookla and M-Lab speed tests) have known limitations, and historic public broadband investments ($42B) are being allocated based on FCC coverage maps that are widely acknowledged to be inaccurate. Comparatively, small investments in improved broadband performance monitoring techniques can help ensure that Internet Service Providers (ISPs) receiving public money to build broadband infrastructure actually deliver on their commitments.
The Challenge
Connecting people affordably to the Internet requires knowing who does not have connection at sufficient speeds and what that insufficient Internet connectivity or alternatives cost. While fixed broadband measurements focus on speed tests over wired (Ethernet) connections, the user experience of home internet is over Wi-Fi. However, the challenge process for consumers who wish to dispute their Internet performance discounts speed tests that are conducted over Wi-Fi citing the unreliability of the measurement. Current methodologies do not contextualize the Wi-Fi environment that the test was conducted over, leading to users being stuck with poor-quality speeds without the technical capabilities to issue an effective challenge.
Proposed Solution
The University of Notre Dame, along with partners ASSIA and the City of South Bend, are developing techniques for addressing the above challenge. Our approach is as follows:
- Through a partnership with the city, w
e have solicited participation from addresses where poor Internet performance has been observed in the past.
- For each household that elects to participate, a Raspberry Pi is provided to test Internet connectivity across a wired (Ethernet) and wireless (WiFi) link, as shown in Figure 1.
- The study seeks to install Raspberry Pis in roughly 50 residential locations across South Bend.
- Each Raspberry Pi is programmed to automatically conduct performance tests (downlink/uplink/latency) multiple times each day measuring performance using a variety of speed test tools including iperf and Ookla over both Ethernet and Wi-Fi to deliver a complete view of the lived Internet experience for those residential users.
- Diagnostic information regarding the WiFi environment is also monitoring examining the extent to which the WiFi may or may not be the bottleneck including aspects such as frequency, bandwidth, RSSI, the number of overlapping APs, number of devices connected, and channel utilization.
- All data is strictly anonymized with only performance data and WiFi environmental data as observed by the Raspberry Pi being saved. Recruitment procedures and incentives were approved by the Notre Dame Institutional Review Board.
- From the observed performance data, an anonymized data repository has been created allowing for data analysis with our partners at ASSIA to study patterns and root causes for poor Internet connectivity.
Team
Professors Monisha Ghosh and Aaron Striegel, grad students Francis Gatsi and Saeid Mehrdad, and Postdoc Muhammad Rochman represent the University of Notre Dame.
Dr. John M. Cioffi, Dr. Peter Chow, Dr. Mehdi Mohseni, and Dr. Peter J. Silverman represent ASSIA.
Patrick McGuire represents the City of South Bend.