The Internet Microscope
As internet services become more integrated into our everyday lives, their share of worldwide electricity consumption has become significant. Existing regional analyses of Internet traffic can give an indication of how energy consumption is distributed between services on aggregated level, but do not allow us to understand the environmental impact of specific internet services, and to what extent people find them valuable in their everyday life [1].
In this project we are developing a standardised way of collecting anonymised data on Internet use that can be scaled up to include a large number of participants. Drawing on previous work on household network monitoring [2,3], we are developing a device that can be mailed to a participating household, plugged into the router for 30 hours by the participant, and then mailed back for analysis. During the monitoring period, participants are asked to fill out time-use diaries. These provide context for the traffic summaries gathered with our network monitoring device, and will aid in the categorization of Internet service usage. Figure 1 below provides an illustration of a “netpack” that a household which has chosen to participate could receive in the mail. We are currently preparing a pilot test with a group of up to 20 households.
More information on how to participate
If you live in Sweden and would like to participate in our study, you can find more information here.
Configuration of the network monitoring device
As part of the study, we evaluated several candidate devices and found that an off-the-shelf (pictured below) travel router is a promising candidate for our network monitoring device. The device comes pre-installed with OpenWrt, a Linux-based operating system adapted for embedded computing, which supports packages for network monitoring (NetFlow). Our selection process also took into account the size of the device, that it would be small enough to fit inside an envelope sent with regular post, and that it should support sufficiently high bandwidth to avoid slowdown of the participants’ connections.
Figure 2. The monitoring device selected for further testing via the initial study. It measures about 12 x 9 x 3 cm. Source: https://www.gl-inet.com/products/gl-mt3000/
The device used for network monitoring, in the picture above, Gl.inet MT3000 wireless router (Beryl AX) with OpenWrt 21.02 operating system. We have installed additional OpenWrt packages to configure it as a network monitoring device: softflowd as softflow sensor; and nfdump as softflow collector. We have also installed bind-dig package to be able to do reverse DNS look-up continuously on the device and openssh-sftp-server package in order to be able to copy data from the device via ssh. Previously, we have tested the more expensive, and more powerful, Gl.inet AXT1800 wireless router (Slate AX) with OpenWrt 21.02 operating system (Kernel version 4.4.60).
If you are a researcher and would like to know more about how to set up an Internet microscope, please contact Mike Hazas. Include a brief description of your intended project, funder, and local ethics procedure.
Key researchers
Jeanelle Carty, Mike Hazas, Kevin Dalli
Related master theses
Viktor Björk, 2024. Temporal Network Traffic Analysis with the Internet Microscope. Civilingenjör IT. https://urn.kb.se/resolve?urn=urn%3Anbn%3Ase%3Auu%3Adiva-532520
Mujtaba Fadhil Jawad, 2024. Self-Supervised Graph Neural Networks Uncover User Activity Patterns. Master in Computer Science (IT dept). https://urn.kb.se/resolve?urn=urn%3Anbn%3Ase%3Auu%3Adiva-540571
Acknowledgement
The Internet Society Foundation funded an initial project "Developing the Internet Microscope" (2021-2024) to develop and pilot test the method. A four-year, follow-on project entitled "The Internet Microscope: Assessing the energy and environmental Impact of online services, and their role in everyday life" has been funded by Formas.
References
[1] Janine Morley, Kelly Widdicks, and Mike Hazas. 2018. Digitalisation, energy and data demand: The impact of Internet traffic on overall and peak electricity consumption. Energy Research & Social Science 38, (April 2018), 128–137. DOI:https://doi.org/10.1016/j.erss.2018.01.018
[2] Kelly Widdicks. 2020. Understanding and mitigating the impact of Internet demand in everyday life. PhD Thesis. Lancaster University. DOI:https://doi.org/10.17635/lancaster/thesis/951
[3] Kelly Widdicks, Mike Hazas, Oliver Bates, and Adrian Friday. 2019. Streaming, Multi-Screens and YouTube: The New (Unsustainable) Ways of Watching in the Home. In Proceedings of the 2019 CHI Conference on Human Factors in Computing Systems, ACM, Glasgow Scotland Uk, 1–13. DOI:https://doi.org/10.1145/3290605.3300696