Evaluation of the impact of different network telemetry collection approaches on machine learning models

• Grant number: 24/03181-0
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Effective date (Start): April 01, 2024
• Status: Discontinued
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computer Systems
• Principal Investigator: Fabio Luciano Verdi
• Grantee: Gabriel Santos de Andrade
• Host Institution: Centro de Ciências em Gestão e Tecnologia (CCGT). Universidade Federal de São Carlos (UFSCAR). Campus de Sorocaba. Sorocaba , SP, Brazil
• Host Company: Universidade Estadual de Campinas (UNICAMP). Faculdade de Engenharia Elétrica e de Computação (FEEC)
• Associated research grant: 21/00199-8 - SMART NEtworks and ServiceS for 2030 (SMARTNESS), AP.PCPE
• Associated scholarship(s): 24/13591-1 - Evaluation of the impact of different network telemetry collection approaches on machine learning models, BE.EP.IC

Abstract

Inserting intelligence into the context of network telemetry using In-band Network Telemetry (INT) is a natural step, as research in the field of computer networks has advanced the use of machine learning (ML) algorithms. In the literature, some works have used ML in conjunction with a programmable data plane and made extensive use of telemetry data collected from the network to feed ML models.However, while the amount of telemetry data collected is, on the one hand, extremely important for ML models, on the other hand it can cause excessive consumption of network resources just to collect such metrics. In this sense, there are several techniques in the literature that use deterministic and probabilistic approaches to reduce the amount of telemetry data collected. However, there are no studies that evaluate how these techniques impact the accuracy of ML models, as less telemetry data will be collected from the network and injected into such algorithms.This project, therefore, proposes to collect network telemetry metrics using different probabilistic and deterministic approaches, evaluating how such approaches affect the accuracy of ML models.