Contacts

Dr. Romain Jacob

Prof. Laurent Vanbever

Today, the information and communication technology (ICT) industry has a massive carbon footprint (a few percent of the worldwide emissions) and one of the fastest growth rates. The Internet accounts for a large part of that footprint while being also energy inefficient. In 2003, some researchers estimated that the core of the Internet consumes between 2 and 24 times more energy per byte than wireless LAN [1]. This inefficiency can be explained by three facts.

1. Network devices are always on;
2. Network devices’ energy consumption is mainly independent of traffic load (see figure);
3. Network devices are under-utilized.

Network operators design their networks to support peak traffic and be resilient to failures. Average utilization between 20-40% are common [2]. Hence, we operate networks in their least energy-efficient regime most of the time…

⛔ From a sustainability perspective, this is unacceptable! ⛔

In this project, we investigate the potential energy savings achievable by turning parts of networks off. For example, there is consistently less traffic during the night; we do not need to leave the entire network on all the time.

Today, network devices’ power draw is dominated by the “idle power,” the power required just to turn them on without forwarding any traffic.

Benefits

With this project, you will

• Contribute to the emerging research activities on sustainable networking;
• Learn how to design and use analytical power models;
• Understand better the trade-off between performance, reliability, and energy efficiency;
• Study a large set of real-world network topology [3] and estimate the potential energy savings of turning devices off.

Prerequisites

For this project, you should be

• Proficient with Python;
• Comfortable with analytical models;
• Knowledgeable about graph theory, or interested in learning;
• Interested in sustainability.

<aside> ♻️ This is only an initial step among a much larger research project that our group is starting [4]. Most likely, there will be follow-up projects for interested students.

</aside>

Having followed our Communication Networks lecture (or equivalent) is a plus, but it is not required.

References

[1] Maruti Gupta and Suresh Singh. Greening of the internet. SIGCOMM ’03.

[2] Cisco. Cisco WAN Automation Engine (WAE) - Best Practices in Core Network Capacity Planning.

[3] The Internet Topology Zoo.

[4] Romain Jacob, Laurent Vanbever. The Internet of tomorrow must sleep more and grow old. HotCarbon’22.