5/6G enhanced adversarial attack defense algorithm using the NS3 simulator

Abstract

This paper presents an enhanced adversarial attack defense algorithm tailored for 5G and 6G wireless networks. The proposed method integrates deep neural networks (DNNs) with real-time network monitoring to detect and mitigate adversarial machine learning attacks, including Fast Gradient Sign Method (FGSM), Projected Gradient Descent (PGD), Basic Iterative Method (BIM), and Momentum Iterative Method (MIM). Using the NS-3 simulator with Reconfigurable Intelligent Surfaces (RIS) and Joint Communication and Sensing (JCAS), the algorithm dynamically adjusts beamforming parameters and threat responses based on evolving traffic and signal conditions. The performance of the defense mechanism is evaluated through simulation, comparing it to a recent benchmark model. Key metrics such as user equipment (UE) localization accuracy, network latency, throughput, and computational overhead are analyzed under various attack and load scenarios. Results show that the proposed solution improves detection effectiveness while maintaining acceptable trade-offs in latency and throughput. This study demonstrates the viability of integrating adaptive learning-based defenses into next-generation networks and provides a practical simulation framework for future research on adversarial robustness in wireless environments.