Document Constraints & Influences

[jlivingood]

The paper suggests that the test is able to estimate aggregate connection speeds but is silent on factors influencing & constraining the measurements. For example, does the test sense whether other users are concurrently using the connection and thereby reducing available per-user or per-device capacity (and not record a result in the presence of competing traffic)? Does the test record whether the device was on WiFi or ethernet and can the dataset be separated based on connection type? For WiFi connections, is the 802.11 protocol, WiFi SNR or number of other clients on the WiFi network recorded in the test, potentially enabling filter out older/slow WiFi standards, devices at the edge of coverage, etc.? What about device type and device capabilities (e.g. do you record if it is a 15 year old laptop that cannot exceed 10 Mbps)?

Reference: https://cacm.acm.org/magazines/2020/12/248801-measuring-internet-speed