OFC2025 Workshop and Technical program presentations & involvement by UvA group members.
OFCnet workshop; Sunday, March 30, 13:00 – 15:30, Level 2 - Rooms 203-204: Networks of the Future and Next-Generation Production.
OFCnet2025’s workshop embodies the OFCnet vision to be
a bridge between the OFC technical conference and “networks of the
future”. Over 2.5 hours, the panel-format workshop is a forum that
enhances development of next-generation networks by highlighting
cutting-edge technologies introduced during the OFC conference that are
gaining maturity and will impact near-term production design as well as
the networks of the future. The workshop focuses on these four key
areas:
Automation and orchestration: evolving the network operations team
Trend analysis and failure prediction: standardized data collection for machine learning
Quantum networking early wins: report-out on the state of implementation
Networks of the future: bleeding-edge technologies and how soon will they be here?
Each area will have two presenters to set the
stage, followed by a moderated panel+audience discussion on the topic to
explore the adoption of these technologies within production networks
and the associated challenges.
Organizers: Sana Bellamine (Lead), CENIC MMBI, United
States, Scotty Strachan (Lead), Nevada System of Higher Education,
United States, Akbar Kara, Ciena, United States,
James Stewart, Utah Education and Telehealth Network, United States, Chris Tracy, ESnet, United States
Speakers
1th panel moderated by Akbar Kara:
James Deaton, Internet2, United States
Hideki Nishizawa, NTT, Japan
2nd panel moderated by Chris Tracy:
Danial Ebling, Utah Education and Telehealth Network, United States
Reza Rokui, Ciena, Canada
John Wu, ESnet, United States
3th panel moderated by Scotty Strachan:
Mariam Kiran, Oak Ridge National Laboratory, United States
Wenji Wu, Energy Sciences Network, United States
4th panel moderated by Sana Bellamine:
Lidia Galdino, Corning Optical Communications, United Kingdom
UvA participates with one demo in the OFCnet booth 5137
In-Band Network Telemetry Based Path Trust.
Every user should be assured that data is routed
securely through a Trusted Path, even in case of congestion or flow
steering. We demonstrate a testbed with optical connections between
programmable P4 switches that utilize In-Band Network Telemetry (INT).
All P4 switches share the same key to encrypt communication with a
Telemetry Collector and to decrypt packet header control information.
All INT packets carry encrypted Path Tracing and Trust information about
their source node. This establishes transparency, route accountability
and trust to the path and the underlying hardware infrastructure. For
more info see below.
UvA presence in OFCnet panel in EXPO II:
OFCnet demonstrations go far beyond the optical data
layer. Research continues to identify applications and opportunities to
benefit from the speed and reliability of optical transport. In this set
of presentations we will hear from researchers who are using the
optical layer as the foundation for next advancements in their areas of
expertise.
Thursday, April 3, 15:15 -15:45
dr. Anestis Dalgkitsis
Digital Sovereignty in Practice: Leveraging In-Band Telemetry and ML for a Responsible Internet
Digital Sovereignty in Practice: Leveraging In-Band Telemetry and ML for a Responsible Internet.
As digital technologies continue
to integrate into governance, commerce, and communication, it is
essential for societies to maintain their autonomy and ensure that
critical systems remain resilient against external manipulation or
surveillance. A responsible Internet should empower not only providers
of critical services but also individuals to access and choose the
equipment handling their data. Users should also have enough control
over the destiny of their data by specifying several requirements, such
as trusted networking equipment and preferred geographical location.
Moreover, it must enable users to verify whether operators act in good
faith and trace incidents or attacks back to their root causes.
Emerging technologies such as
programmable networks, In-band Network Telemetry (INT), Machine Learning
(ML), and Intent Based Networking (IBN) play a pivotal role in
realizing these goals. The advent of Programmable Data Planes (PDPs) has
enabled INT, which offers significant advantages, including flexible
programmability and real-time, detailed network visibility. This is
achieved by embedding network state information directly into data
packet headers, allowing the data plane to independently manage network
measurements without intervention from the control plane. Driven by the
socioeconomic outlook for the future of the Internet, we have designed,
developed, and demonstrated a Responsible Internet proof-of-concept
(PoC). Our work contributes to the field in three key ways, by:
Enabling users to intuitively specify and
monitor the path of their data on the Internet through an operator
platform, and verify the trustworthiness of the path by using INT.
Implementing a Reinforcement Learning (RL)
approach in programmable devices that autonomously learns and selects
optimal, secure routes based on INT-collected metrics and user
preferences, facilitating dynamic path optimization directly in the data
plane without control plane intervention.
Conducting realistic experiments on the FABRIC
network infrastructure, in order to validate the feasibility and
practicality of this PoC.
