Developing the Internet Microscope

Research Question

1. How does the traffic volume and energy requirement of Internet services support what people do in everyday life? Conversely, what kinds of traffic are difficult to tie to any value or meaning – and thus might be wasteful?

Why Is This Research Important?

Our understanding of the environmental impact of digital services relies on regional traffic reports and broad “top-down” estimates of network and data center footprints. A crucial missing piece is a “bottom-up” perspective. This research is vital because it seeks to link specific Internet services used in households to the amount of traffic they generate, their energy footprint, and ultimately, how they relate to household activities.

Making this connection is essential for companies, organizations, and governments to develop effective policies that support a sustainable Internet, one that genuinely enhances the most important and meaningful aspects of everyday life. It can increase awareness and reflection for individuals about the environmental impact of their online activities, empowering them to reduce waste and have more meaningful digital experiences.

Methodology

The project employed an adapted sampling method using a “NetPack” sensor kit, which was delivered to households. The NetPack contained several components:

– A network data logger: This device sits between the household’s broadband modem and the WiFi router to monitor network traffic.

– A pulse counter: This device attaches to the household electricity meter to log real-time energy consumption.

– Mobile phones are provided for every adult in the household. These phones run a time-use diary app. The time-use diary app is based on established methodologies, such as those used in the UK.

– Time Use Survey and the Multinational Time Use Study. Participants use the app to log their activities and the digital devices associated with them.

Each sampled household participated for a thirty-hour window. While the NetPacks primarily captured usage on home networks (not cellular), the time-use diaries provided some indication of device use during activities., Analysis using data gathered from households via the NetPacks was directly used to estimate carbon emissions and energy consumption.

Media and Publications

• The Internet Microscope Project website hosting all the community resources and research outputs
• Workbook to help participants reflect on their use of online services
• The Internet Microscope: Scaling the Study of Online Services in the Home Analysis of data gathered from households via the NetPacks used to estimate carbon emissions and energy consumption.
• Temporal Network Traffic Analysis with the Internet Microscope Study evaluating the potential for classifying NetFlow data from a network where one service is being used at a time.
• Self-Supervised Graph Neural Networks Uncover User Activity Patterns This work explores algorithms for self-supervised learning in order to automatically cluster different flows into different services.