Session

Artificial Intelligence and Machine Learning in Networking

Chairs

Roopa Prabhu
Marta Plantykow

Label

Nuts and Bolts

Session Type

Workshop

Contents

Description

In this workshop we have a lineup of talks and tutorials on Artificial Intelligence and Machine learning (AI/ML) algorithms and techniques applied to Networking in the areas of optimizations, security, operations and more.

Talk list: (a) We will start with a tutorial by Marta Plantykow m.plantykow@gmail.com “Bridging networking and AI: Empowering Networking Engineers with Artificial Intelligence”: Overview of the fundamental concepts of AI/ML and their relevance to networking. Through relatable examples and approachable explanations, attendees will better understand how AI can enhance networking. While the discourse maintains a high-level perspective, it aims to demystify the notion that AI is overly complex and inaccessible.

Key takeaways from this tutorial will include:

  • An overview of the fundamental concepts of AI/ML and their relevance to networking engineering
  • Illustrative use cases showcasing how AI-driven applications in networking

(b) Genetic Algorithm based PI controller tuning with stability analysis for Linux ptp4l optimization

This talk presents the research results of ptp4l Proportional-Integral (PI) controller tuning using a Genetic Algorithm. Furthermore, it describes recent advancements in the PTP-optimization framework. PI controller optimization is required to meet the strict phase synchronization industry standards introduced by the LTE E-ULTRA TDD, 5G, and O-RAN technologies. The ptp4l is a part of the linuxptp toolbox and implements Precision Time Protocol (PTP) according to IEEE standard 1588 for Linux. The tool implements a PI controller and an API to tune its Proportional (P) and Integral (I) terms. The developed framework implements Genetic Algorithm (GA), a stochastic algorithm used in Artificial Intelligence (AI), to optimize the parameters of the PI controller. Stability condition checks are incorporated into the framework to accelerate the discovery of optimal and stable outcomes. The enhancements resulting from utilizing the tool are presented in the article. This paper continues research for tuning the Proportional-Integral (PI) controller configurable in the phc2sys application using Genetic Algorithm presented on Netdev 0x15.

(c) A generic ML-enhanced controller that optimizes network application efficiency by automatically controlling performance and energy trade-offs - Han Dong

(d) Towards an Autonomous Data Center - Hanan Shteingart

(e) Generative AI and LLMs in networking and security - Roopa Prabhu