The More the Better: Feeding Machine Learning Algorithms with INT Data

Federal University of São Carlos (UFSCar) - LERIS - Laboratory of Network, Innovation and Software

This project will use in-band network telemetry data to feed (as much as possible) Machine Learning (ML) algorithms to support service assurance. The idea is to put together network and service metrics to attend several aspects such as pro-active reaction and autonomous healing. The collection of INT data will be done using a setup formed by one NetFPGA SUME card and 5 Netronome Agilio CX Dual-Port 10 Gigabit Ethernet SmartNICs. The main idea is to detect at line rate long queues formation in the dataplane so that a make-before-brake approach is used to keep the QoE of e2e applications. 

Fábio Luciano Verdi

Other collaborators: Leandro Almeida