Generating Practical Adversarial Network Traffic Flows Using NIDSGAN

Abstract

Network intrusion detection systems (NIDS) are an essential defense for computer networks and the hosts within them. Machine learning (ML) nowadays predominantly serves as the basis for NIDS decision making, where models are tuned to reduce false alarms, increase detection rates, and detect known and unknown attacks. At the same time, ML models have been found to be vulnerable to adversarial examples that undermine the downstream task. In this work, we ask the practical question of whether real-world MLbased NIDS can be circumvented by crafted adversarial flows, and if so, how can they be created. We develop the generative adversarial network (GAN)-based attack algorithm NIDSGAN and evaluate its effectiveness against realistic ML-based NIDS. Two main challenges arise for generating adversarial network traffic flows: (1) the network features must obey the constraints of the domain (i.e., represent realistic network behavior), and (2) the adversary must learn the decision behavior of the target NIDS without knowing its model internals (e.g., architecture and meta-parameters) and training data. Despite these challenges, the NIDSGAN algorithm generates highly realistic adversarial traffic flows that evade ML-based NIDS. We evaluate our attack algorithm against two state-of-the-art DNNbased NIDS in whitebox, blackbox, and restricted-blackbox threat models and achieve success rates which are on average 99%, 85%, and 70%, respectively. We also show that our attack algorithm can evade NIDS based on classical ML models including logistic regression, SVM, decision trees and KNNs, with a success rate of 70% on average. Our results demonstrate that deploying ML-based NIDS

Resilient intrusion detection system for adversarial attacks on Low-Rate DDoS

“…After conducting experiments, it was necessary to compare our results with the results of existing studies and models to show the effectiveness of our model. [26] and [28]. Compared with [18], we found some results close to our results, especially when using Genetic Attack.…”

“…For more evasive attacks, the authors in [28] proposed NIDSGAN, a GAN-based attack method, then added new terms to the loss function. The loss function decreases the distance between the real and adversarial features and the conventional loss function, which prevents the discriminator from distinguishing between adversarial and real traffic.…”

Towards Real-time Network Intrusion Detection with Image-based Sequential Packets Representation

“…It changes the following bytes from the IP header: fragmentation bytes (byte number 7 and 8), TTL byte (byte number 9), and IP checksum bytes (byte number 11 and 12). To compare the performance of our approach, we tested the perturbed samples against the packet and flow-based NIDS used in recent literature studies [51], [50], which included decision tree, random forest, support vector machine, k-nearest neighbor, and deep neural network models. The evasion rate (ER) against these ML/DL-based NIDS ranged from 70% to 99% across different attack types.…”

StinAttack: A Lightweight and Effective Adversarial Attack Simulation to Ensemble IDSs for Satellite- Terrestrial Integrated Network

“…A few researchers began employing simple deep learning models to detect malicious traffic. Such as, AlertNet [26], DeepNet [27], and IdsNet [28] are based on fully connected perceptrons with ReLU activation functions, batch normalization to improve training performance, and dropout to prevent overfitting. For non-differentiable models, we primarily use the decision trees (DT), Random forests (RF), XGboost (Xgb.…”