CBL - Campus del Baix Llobregat

Projecte llegit

Títol: Synthetic traffic generation with Artificial Intelligence

Estudiants que han llegit aquest projecte:

Director/a: GARCÍA VILLEGAS, EDUARD

Departament: ENTEL

Títol: Synthetic traffic generation with Artificial Intelligence

Data inici oferta: 28-06-2023     Data finalització oferta: 28-02-2024


Estudis d'assignació del projecte:
    DG ENG AERO/SIS TEL
    DG ENG AERO/TELEMÀT
    DG ENG SISTE/TELEMÀT
Tipus: Individual
 
Lloc de realització: EETAC
 
Paraules clau:
Machine Learning, Artificial Intelligence, NS-3, Synthetic traffic
 
Descripció del contingut i pla d'activitats:
The generation of synthetic traffic has many applications. In network security, intrusion detection systems or firewalls must be trained to detect malicious traffic by studying its patterns. Training those systems with real devices in a real environment can be costly and has risks. In network simulation environments, the traffic sources are often overly simplistic, making it difficult to assess the performance of different protocols under study accurately.

Alternatively, synthetic traffic sources can be used to train those network security systems offline or to generate realistic traffic within a simulated environment. Those traffic sources should be capable of mimicking the behavior of real traffic as much as possible.

In this project, the student will work with a pre-existing Python framework to generate synthetic IoT traffic, based on processing real-network traces with machine learning techniques. The traffic generated will be analyzed through NS-3 simulations. The results of this work will be published in a scientific journal.
 
Overview (resum en anglès):
Artificial intelligence (AI) has become a transformative force in today's world, revolutionising industries by enabling smarter decision-making and automation. Its ability to analyse vast datasets, learn from patterns, and adapt to new challenges makes it essential for solving complex problems in areas like cybersecurity, healthcare, and communications. As networks and systems grow in complexity, AI plays a critical role in enhancing efficiency, improving security, and driving innovation across various fields.
This project focuses on three key objectives: first, to explore artificial intelligence, with a deeper emphasis on machine learning techniques, and their applicability in the context of traffic generation. AI and ML models have shown great promise in analysing complex datasets, making them suitable for modelling the behaviour of real-world traffic and generating synthetic traffic that mimics real-world conditions. A thorough analysis of various ML algorithms will be conducted, examining how they can be applied to predict traffic patterns and generate synthetic data that aligns with the characteristics of real IoT traffic.
The second component of the project involves capturing real-world traffic from two specific mobile games. Mobile gaming is a growing area where fast and stable communication is crucial, with constant data exchange between the game servers and players' devices. By analysing the traffic from these games, the project will gain insights into the patterns, volume, and nature of data transmitted in real-time, providing a solid foundation for modelling synthetic traffic. This real-world traffic captured will form the basis for developing a realistic dataset, which will be instrumental in generating synthetic traffic using AI-based approaches.
Finally, the project will use the captured data to generate synthetic traffic and assess its accuracy and realism compared to the actual traffic data. The goal is to determine how closely the synthetic traffic matches the real traffic in terms of key characteristics, such as transmission frequency, data volume, and event-based spikes. A comprehensive evaluation will be performed to measure the quality of the generated traffic, identifying strengths and potential areas for improvement.

© CBLTIC Campus del Baix Llobregat - UPC