Innovative approaches for network analysis and optimization: leveraging deep learning and programmable hardware

• Grant number: 24/05363-9
• Support Opportunities: Scholarships in Brazil - Doctorate
• Start date: May 01, 2024
• Status: Discontinued
• Field of knowledge: Engineering - Electrical Engineering - Telecommunications
• Principal Investigator: Christian Rodolfo Esteve Rothenberg
• Grantee: Ariel Góes de Castro
• Host Institution: Faculdade de Engenharia Elétrica e de Computação (FEEC). Universidade Estadual de Campinas (UNICAMP). Campinas , SP, Brazil
• 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): 24/22594-4 - Deep Learning-Driven Traffic and Trace Generation for Beyond 5G Network Architectures and Services, BE.EP.DR

Abstract

The surge in network demand fueled by real-time, data-intensive applications like healthcare, autonomous vehicles, and cloud gaming has intensified the need for enhanced network performance. However, production networks often suffer from issues like latency and congestion, impacting user experience and service quality. To effectively develop new algorithms and solutions that cater to the diverse requirements of such applications under realistic conditions, it is imperative to test them in environments that closely resemble real-world scenarios. This necessitates using real devices and incorporating communication logs. However, a crucial challenge lies in the limited availability of freely accessible network traces for academic research. This scarcity hinders the ability to comprehend the needs of various applications under realistic network conditions. In this context, we propose leveraging the power of neural networks. These networks possess the remarkable ability to learn from existing network traces and gain insights into the characteristics of anomalous events, such as link failures. This knowledge can be employed to augment existing traces with synthetically generated data, carefully considering privacy concerns. By utilizing the augmented traces with synthesized anomalies, we can, for instance, empower routing algorithms designed to benefit from programmable hardware (e.g., SmartNICs) and collected data plane metrics, paving the way for improved network performance and enhanced user experience with more autonomous decisions. (AU)