Session Date/Time: 27 Mar 2023 04:00

irtfopen

Summary

This meeting served as an opening session for the IRTF, featuring updates from the chair, introductions to two proposed research groups (usable formal methods and research and analysis of standard setting processes), and presentations from the winners of the Applied Networking Research Prize. The meeting covered key aspects of the IRTF's ongoing work, including research group activities, publications, and workshops.

Key Discussion Points

IRTF Structure and Focus: The IRTF functions as a research-oriented organization parallel to the IETF, focusing on longer-term research issues affecting the internet, rather than standards development.
New Research Groups:
- Usable Formal Methods (UFM): Aims to bridge the gap between academia and the standards community regarding formal methods for protocol specification, with a focus on improving protocol correctness and usability of formal methods tools.
- Research and Analysis of Standard Setting Processes (RAS): Focuses on understanding the dynamics of standard-setting processes, including community diversity, decision-making, and interactions between participants.
Applied Networking Research Prize: Awarded to recognize significant contributions to applied networking research with potential relevance to internet standards.
- Boris as many: Novel offload architectures for NICs
- After seller jacob: Machine learning for network security
Autonomous NICs: Boris as many presented work on autonomous NICs, describing a software/hardware architecture that offloads L5 protocol computations (e.g., TLS encryption) to the NIC while maintaining software control of the network stack. The discussion covered the design space for L5 protocol acceleration and the benefits and limitations of different approaches, including trade offs with memory use.
AI/ML for Network Security: After seller jacob presented work on explainable AI for network security, specifically the "Trustee" framework for generating global explanations of machine learning models in the form of decision trees. The discussion emphasized the importance of understanding why a model works and when it doesn't. Several use cases revealed issues with commonly used datasets, including data leakage, overfitting, and spurious correlations. The importance of high quality datasets was emphasized.

Decisions and Action Items

Attendees were encouraged to attend the introductory meetings of the Usable Formal Methods research group (Wednesday) and the Research and Analysis of Standard Setting Processes research group (Thursday).
Attendees were encouraged to submit papers to the Applied Networking Research Workshop by the deadline (May 12th).
Attendees were encouraged to apply for travel grants to attend IRTF meetings and the Applied Networking Research Workshop (call for travel grants expected later this month).

Next Steps

Follow up on action items.
Continue discussions within the newly proposed research groups.
Prepare for the Applied Networking Research Workshop.