L’Ecole doctorale : Ecole Doctorale de l’Institut Polytechnique de Paris
et le Laboratoire de recherche SAMOVAR – Services rĂ©partis, Architectures, ModĂ©lisation, Validation, Administration des RĂ©seaux
présentent
lâAVIS DE SOUTENANCE de Monsieur Hamdi FRIJI
AutorisĂ© Ă prĂ©senter ses travaux en vue de lâobtention du Doctorat de l’Institut Polytechnique de Paris, prĂ©parĂ© Ă TĂ©lĂ©com SudParis en :
Informatique
« DĂ©tection d’intrusion basĂ©e sur les Graph Neural Networks pour la sĂ©curisation des rĂ©seaux en pĂ©riphĂ©rie »
le MARDI 10 DĂ©CEMBRE 2024 Ă 14h00
Ă
Amphithéùtre 03
19 place Marguerite Perey – 91120 PALAISEAU
et sur le lien zoom:
https://telecom-paris.zoom.us/j/93441814530?pwd=Mqb8aFD3v1X2B9a1VA5tqITxntj0tu.1
ID de réunion: 934 4181 4530
Code secret: 361629
Membres du jury :
M. FrĂ©dĂ©ric LEHMANN, Full professor, Institut Polytechnique de Paris , FRANCE – Directeur de these
M. Erol GELENBE, Full professor, Institute of Theoretical and Applied Informatics, Polish Academy of Sciences, POLOGNE – Rapporteur
M. Imed ROMDHANI, Associate Professor, Edinburgh Napier University, ROYAUME-UNI – Rapporteur
M. Fabrice ROSSI, Full professor, UniversitĂ© Paris Dauphine, FRANCE – Rapporteur
M. Pierre-Francois GIMENEZ, Docteur, INRIA, FRANCE – Examinateur
M. Maxime LABONNE, Docteur, Liquid AI, ETATS-UNIS – Examinateur
Mme Mireille SARKISS, MaĂźtre de confĂ©rences, Institut Polytechnique de Paris, FRANCE – Co-encadrant de these
M. Rida KHATOUN, Full professor, Institut Polytechnique de Paris, FRANCE – Examinateur
Invité :
M. Alexis OLIVEREAU, Ingénieur de Recherche, CEA-LIST, Encadrant de thÚse
« DĂ©tection d’intrusion basĂ©e sur les Graph Neural Networks pour la sĂ©curisation des rĂ©seaux en pĂ©riphĂ©rie »
présenté par Monsieur Hamdi FRIJI
Résumé :
Face Ă l’escalade de la complexitĂ© et Ă la frĂ©quence des cyberattaques, cette thĂšse propose des approches innovantes pour la dĂ©tection d’intrusion dans les rĂ©seaux, en exploitant les capacitĂ©s avancĂ©es des rĂ©seaux de neurones en graphe (Graph Neural Networks, GNNs) et de nouvelles reprĂ©sentations sous forme de graphes. Nous commençons par une analyse critique des jeux de donnĂ©es et des reprĂ©sentations de rĂ©seaux actuels, en abordant des questions clĂ©s sur leur efficacitĂ©. Nous introduisons une nouvelle reprĂ©sentation des flux de communication sous forme de graphes, offrant une plus grande robustesse face aux manipulations de type attaques adversariales (adversarial attacks). Nous prĂ©sentons ensuite lâun des premiers systĂšmes de dĂ©tection d’intrusion utilisant notre reprĂ©sentation en graphe basĂ©e sur les GNNs. Ce systĂšme permet d’Ă©valuer les comportements malveillants en capturant des motifs complexes souvent ignorĂ©s par les mĂ©thodes traditionnelles. Les rĂ©sultats montrent que notre approche surpasse largement les solutions existantes basĂ©es sur l’apprentissage automatique et les GNNs en termes de prĂ©cision et de robustesse. Pour relever les dĂ©fis de la scalabilitĂ© et de l’efficacitĂ© dans les environnements Ă grande Ă©chelle, nous introduisons G-DEMIS (« Graph-based DEcentralized Multi-agent Intrusion detection System »), un systĂšme multi-agent dĂ©centralisĂ© exploitant les GNNs pour une dĂ©tection rapide des activitĂ©s malveillantes. En agrĂ©geant les informations locales Ă travers le rĂ©seau, G-DEMIS amĂ©liore la dĂ©tection en temps rĂ©el tout en rĂ©duisant la consommation d’Ă©nergie de 58 % et le temps de rĂ©ponse de 17,13 % par rapport aux approches centralisĂ©es. Enfin, nous proposons un algorithme innovant pour tracer les chemins de propagation des attaques, aidant Ă identifier les machines compromises dans le contexte des menaces persistantes avancĂ©es. Les travaux de cette thĂšse permettent non seulement de faire progresser l’Ă©tat de l’art en matiĂšre de dĂ©tection d’intrusions, mais ouvrent Ă©galement la voie Ă de nouvelles avancĂ©es en cybersĂ©curitĂ©.
Abstract :
In light of the escalating complexity and frequency of cyberattacks, this thesis presents innovative approachs to network intrusion detection that leverages the advanced capabilities of Graph Neural Networks (GNNs) and novel graph-based representations. To lay the foundation for our research, we first conduct a critical review of existing intrusion detection datasets and network representations, focusing on their effectiveness in addressing key research challenges. This thesis presents our insights and analysis of two widely used datasets: ToN IoT and CICIDS 2017, highlighting their strengths and limitations. Our approach introduces a new flow-based graph representation of communication flows, which enhances existing solutions by increasing robustness against adversarial attacks. First, we present one of the pioneering GNN-based intrusion detection systems, which utilizes our graph representation and GNN algorithms to compute maliciousness scores. This system captures complex relational patterns that traditional methods often overlook. Our findings demonstrate that this framework significantly outperforms the current state-of-the-art machine learning and GNN-based solutions in terms of both accuracy and robustness. Additionally, we propose a three-stage intrusion detection system inspired by the Lockheed Martin cyber kill chain, designed to detect advanced multi-step attacks. This system achieved an average F1-score of 94% on the ToN IoT dataset, surpassing traditional random forest models and demonstrating its effectiveness for real-world applications. To address scalability and efficiency challenges in large-scale environments, we introduce G-DEMIS, a Graph-based DEcentralized Multi-agent Intrusion detection System that enhances the use of GNNs for a fast detection of malicious activities. G-DEMIS employs a collaborative approach in which multiple agents monitor different network segments, aggregating local graph information to form a comprehensive view of the network. This framework not only enhances real-time detection capabilities but also reduces energy consumption by 58.08% and detection time by 17.13% compared to centralized models. Finally, we tackle the challenge of defending against Advanced Persistent Threats (APTs) by proposing a novel algorithm for reconstructing attack propagation paths. This algorithm assists engineers in identifying compromised machines following an APT attack by detecting and analyzing anomalous behaviors in the network, tracing the progression of the attack, and providing a detailed understanding of the attack paths. This thesis not only advances the current state of intrusion detection but also lays the foundation for future innovations in cybersecurity.