Improving Network Energy Consumption while Maintaining Error-Resilience

Fri, 01 May 2026 00:00:00 +0000

Background

ISP networks consist of thousands of routers and network elements interconnected through numerous links. Each of these components consumes significant amounts of energy, making ISPs one of the major global energy consumers. Mitigating corresponding environmental and sustainability impacts is among the most pressing needs of society.

One particular challenge in this space is that the energy consumption of today’s networking hardware is relatively independent of traffic load. As a consequence, energy usage remains similar across the day although traffic load can vary by more than 60%. However, this also means that solutions scaling energy usage with traffic load could present large savings.

One specific solution space is motivated by how ISP networks are often designed: with redundancies and spare capacities in mind to provide error resilience when parts of the network become temporarily unavailable. Hence, it might be feasible to switch off links, line-cards, or even whole routers to reduce energy consumption without negatively affecting the network. However, it is unclear to what extent this is possible and where the limits lie between optimizing energy consumption and sacrificing error resilience and reliability.

Your Challenge

The main goal of this thesis is to explore the sketched solution space and investigate the feasibility and effectiveness of saving energy in ISP networks through shutting down parts of the networks while still complying with hard error-resilience constraints and other practical requirements. Addressing this goal involves multiple related and challenging research questions that we aim to address together with you and our partners from Deutsche Telekom (DT) Technik. A potential research agenda could include:

designing & evaluating energy-optimization algorithms that account for real-world constraints.
studying the limits of pruning the networks by shutting off specific segments in light of practical redundancy requirements.
studying the impact of different time horizons for network pruning to derive practical guidelines for how to best deploy such algorithms.
studying how oscillations in these algorithms can be avoided and whether changes to the underlying network topology could even enable additional improvements, providing valuable insights for network topology designs.

What we offer

The opportunity to address a pressing societal challenge
Collaboration with experts from DT Technik
Access to real-world data from the Deutsche Telekom ISP network

What we expect

Strong interest in large-scale networks and data crunching
Basic knowledge of combinatorial optimization / operations research
Solid programming skills (knowledge of Python + Gurobi is beneficial)
Ability to work independently and communicate effectively and regularly

Green Networking | COMSYS | RWTH Aachen University

Improving Network Energy Consumption while Maintaining Error-Resilience

Background

Your Challenge

What we offer

What we expect