Better Safe Than Never: A Survey on Adversarial Machine Learning Applications towards IoT Environment

Abstract

Internet of Things (IoT) technologies serve as a backbone of cutting-edge intelligent systems. Machine Learning (ML) paradigms have been adopted within IoT environments to exploit their capabilities to mine complex patterns. Despite the reported promising results, ML-based solutions exhibit several security vulnerabilities and threats. Specifically, Adversarial Machine Learning (AML) attacks can drastically impact the performance of ML models. It also represents a promising research field that typically promotes novel techniques to generate and/or defend against Adversarial Examples (AE) attacks. In this work, a comprehensive survey on AML attack and defense techniques is conducted for the years 2018–2022. The article investigates the employment of AML techniques to enhance intrusion detection performance within the IoT context. Additionally, it depicts relevant challenges that researchers aim to overcome to implement proper IoT-based security solutions. Thus, this survey aims to contribute to the literature by investigating the application of AML concepts within the IoT context. An extensive review of the current research trends of AML within IoT networks is presented. A conclusion is reached where several findings are reported including a shortage of defense mechanisms investigations, a lack of tailored IoT-based solutions, and the applicability of the existing mechanisms in both attack and defense scenarios.

RobEns: Robust Ensemble Adversarial Machine Learning Framework for Securing IoT Traffic

“…Consequently, conventional ML-based approaches have been favored for such applications due to their simplicity, stability, and robustness [ 2 , 3 ]. In particular, the research approaches of conventional ML for IDSs in the IoT can be grouped into tree-based, clustering-based, probabilistic-based, and non-probabilistic-based categories.…”

“…In other words, pre-processing techniques are required to ensure ultimate performance and reliable detection for IDSs [ 4 ]. As such, the initial evaluations carried out in [ 2 , 4 ] were extended in this study to develop a framework with adversarial strategies to offer cost-effective means.…”

“…This increases attackers chances to evade IDS and perform “successful” attacks. Such attack scenarios can be categorized into: (i) white-box attacks, where the adversary performs attacks on a known learning process of the target model; and (ii) black-box attacks, where the adversary performs attacks on an unknown learning process of the target model [ 1 , 2 ].…”

“…Moreover, attacks within Internet of Things (IoT) environments that lack well-tailored security solutions proved to have substantial impacts. Intrusion Detection Systems (IDSs) have played an important role in ensuring IoT network security, subject to capacity and processing limitations [ 1 , 2 ]. In addition, Machine Learning (ML) techniques have been widely utilized to enhance the capabilities of several computer applications, including IDSs [ 3 ].…”

“…In terms of attacks, gradient-based methods have been widely adopted because of their strength and the promising results they yield. Similarly, Particle Swarm Optimization (PSO)-based traffic mutation algorithms and GANs [ 2 , 3 ] have been coupled with AML attack strategies. With regards to defense, only a few generic defense techniques, such as adversarial training, have been introduced.…”