VERS UNE ARCHITECTURE DE SÉCURITÉ MULTICOUCHE POUR LA PROTECTION DES RÉSEAUX SANS FIL : ANALYSE ET CLASSIFICATION DES ATTAQUES AVEC MÉCANISMES DE DÉTECTION BASÉS SUR L'APPRENTISSAGE AUTOMATIQUE

Abstract

Wireless Local Area Network (WLAN) security has become a major challenge with the rapid growth of connected devices and wireless access points. The inherent vulnerabilities of radio communications expose organizations to a wide range of threats. This article presents an analysis of WLAN-specific vulnerabilities and a multilayer security architecture based on the defense-in-depth principle. The proposed approach combines several complementary mechanisms: WPA3-Enterprise, 802.1X/EAP-TLS authentication through a RADIUS server, dynamic VLAN segmentation, network access control, an intrusion detection and prevention system based on a hybrid CNN-LSTM model, and an application layer requiring mandatory VPN use. Experiments conducted on a representative infrastructure with 150 attack attempts show an overall detection rate of 95% for the main attack vectors considered. The average time between detection and isolation remains below 150 ms. This study therefore provides a practical methodology for strengthening WLAN security, including in highly constrained environments such as sub-Saharan Africa.