“Self-Adjusting Networks: From Metrics to Algorithms”

Speaker : Stefan Schmid
University of Vienna
Date: 13/01/2021
Time: 2:00 pm - 3:00 pm


In this talk I will present the vision of self-adjusting networks: communication networks whose physical topology adapts to the traffic pattern it serves, in a demand-aware manner. This vision is reminiscent of biased and self-adjusting datastructures, such as Mehlhorn trees and splay trees. Self-adjusting networks are enabled by emerging reconfigurable optical technologies. I will show that the benefit of self-adjusting networks depends on the amount of “structure” there is in the demand, and present an information-theoretical approach to measure the complexity of traffic traces and derive entropy-based metrics accordingly. I will also present optimal offline and online algorithms to design self-adjusting networks whose performance matches the derived metrics asymptotically.

The talk is primarily based on the following papers:

ReNets: Statically-Optimal Demand-Aware Networks
Chen Avin and Stefan Schmid.
SIAM Symposium on Algorithmic Principles of Computer Systems (APOCS), Alexandria, Virgina, USA, January 2021.

On the Complexity of Traffic Traces and Implications
Chen Avin, Manya Ghobadi, Chen Griner, and Stefan Schmid.
ACM SIGMETRICS, Boston, Massachusetts, USA, June 2020.

Demand-Aware Network Designs of Bounded Degree
Chen Avin, Kaushik Mondal, and Stefan Schmid.
Distributed Computing (DIST), Springer, 2020.

SplayNet: Towards Locally Self-Adjusting Networks
Stefan Schmid, Chen Avin, Christian Scheideler, Michael Borokhovich, Bernhard Haeupler, and Zvi Lotker.
IEEE/ACM Transactions on Networking (TON), Volume 24, Issue 3, 2016.

For more details, see the project website:

* * * * * * * * * * * * * * * * * * * * *

Here’s the streaming link: https://telecom-paris.zoom.us/j/94054078947?pwd=VFhyTjIyRWZVYkNzVXp4OXV4c2tTUT09

  • ID de réunion : 940 5407 8947
  • Code secret : 363501