Mixture GAN For Modulation Classification Resiliency Against Adversarial Attacks

Shtaiwi, Eyad; Ouadrhiri, Ahmed El; Moradikia, Majid; Sultana, Salma; Abdelhadi, Ahmed; Han, Zhu

doi:10.1109/globecom48099.2022.10001138

Cited by 7 publications

(5 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The experimental results then show that the proposed GAN network can preserve each modulation type's global structure and restore up to 50% missing samples in the time domain. Differently, the authors in [54] propose a GAN-based modulation classification approach that is resilient to adversarial attacks. Specifically, the authors indicate that conventional DL-based automatic modulation recognition methods are vulnerable to adversarial attacks with well-designed perturbation injected into wireless channels.…”

Section: A Modulation and Signal Classificationmentioning

confidence: 99%

Effect of heavy metals on duckweed growth

Nguyen¹,

Hoang²,

Pham³

et al. 2022

CTUJS

View full text Add to dashboard Cite

In this study, three clones of the duckweed genus Lemna, collected from three different regions of Vietnam Binh Thuan Bac Giang and Ben Tre were used to evaluate the growth and development in a culture medium containing different concentrations of As3+, Cd2+, or Pb2+. During 14 days of incubation, the growth rate was determined daily by measuring the surface area of all plants of each sample. Experimental results show that the three Lemna clones could grow in nutrient medium contaminated by heavy metal at a concentration of 0.3-0.5 mg/L for As3+, 0.15-0.3 mg/L for Cd2+, and 0.15 mg/L for Pb2+. At these concentrations, the division rate of the treated clones was several times higher than that of the controls. The highest increase in the surface area was recorded for Lemna BTN (68.47 times higher when cultured in medium supplemented in 0.3 mg/L Pb2+ than under control conditions). The surface area of Lemna BTR cultured in a medium supplemented with 0.3 mg/L As3+ was 54.65 times higher than in medium without arsenic. The obtained results showed that Lemna BGG and Lemna BTR can tolerate arsenic and cadmium pollution while Lemna BTN tolerates lead pollution.

show abstract

Section: A Modulation and Signal Classificationmentioning

confidence: 99%

Effect of heavy metals on duckweed growth

Nguyen¹,

Hoang²,

Pham³

et al. 2022

CTUJS

View full text Add to dashboard Cite

show abstract

“…The work in [Shtaiwi et al (2022)] proposes a defense technique that discards adversarial samples before they are sent to the modulation classifier. It relies on mixture generative adversarial networks (MGAN) and trains a GAN for each modulation scheme considered.…”

Section: Related Workmentioning

confidence: 99%

“…It relies on mixture generative adversarial networks (MGAN) and trains a GAN for each modulation scheme considered. However, the technique proposed in [Shtaiwi et al (2022)] also significantly increases the computational resources required as one GAN is trained for each modulation scheme. Moreover, the authors of [Shtaiwi et al (2022)] evaluate their proposal against only adversarial samples that are crafted using the FGSM technique and do not indicate the size of adversarial perturbations or if they are imperceptible.…”

Section: Related Workmentioning

confidence: 99%

Towards Intelligent Security Mechanisms for Connected Things

Araujo-Filho,

Campelo,

Kaddoum

2024

Anais Do XXXVII Concurso De Teses E Dissertações (CTD 2024)

View full text Add to dashboard Cite

The widespread adoption of connected devices and the adoption of machine learning enable attackers to launch several cyber-attacks and adversarial attacks. Therefore, the goals of this thesis are to investigate and develop cutting-edge solutions to enhance the security of systems by effectively and efficiently detecting cyber-attacks while also defending systems that rely on ML from adversarial attacks. The main results of our thesis comprehend multiple awards, the publication of eight papers in prestigious journals, three conference papers, two patents, and one software registration. Furthermore, our research has been recognized and awarded as one of the two 2022 Microsoft Research Ph.D. Fellowship recipients in Security, Privacy, and Cryptography worldwide.

show abstract

“…The capsule network Lack of temporal continuity features Shtaiwi 𝑒𝑡 𝑎𝑙. [12] Non-public dataset The generative adversarial network Defending against adversarial attacks Cai 𝑒𝑡 𝑎𝑙. [ solve this problem, Transformers [13] have been proposed recently.…”

Section: Introductionmentioning

confidence: 99%

“…Shtaiwi 𝑒𝑡 𝑎𝑙. proposed a novel AMR method based on generative adversarial network (GAN) to improve the capability of communication systems for adversarial attacks [12]. However, it is mainly targeted at improving the model's ability to defend against adversarial attacks.…”

Section: Introductionmentioning

confidence: 99%

ClST: A Convolutional Transformer Framework for Automatic Modulation Recognition by Knowledge Distillation

Hou,

Li,

Lin

et al. 2024

IEEE Trans. Wireless Commun.

View full text Add to dashboard Cite

With the rapid development of deep learning (DL) in recent years, automatic modulation recognition (AMR) with DL has achieved high accuracy. However, insufficient training signal data in complicated channel environments and large-scale DL models are critical factors that make DL methods difficult to deploy in practice. Aiming to these problems, we propose a novel neural network named convolution-linked signal transformer (ClST) and a novel knowledge distillation method named signal knowledge distillation (SKD). The ClST is accomplished through three primary modifications: a hierarchy of transformer containing convolution, a novel attention mechanism named parallel spatial-channel attention (PSCA) mechanism and a novel convolutional transformer block named convolution-transformer projection (CTP) to leverage a convolutional projection. The SKD is a knowledge distillation method to effectively reduce the parameters and complexity of neural networks. We train two lightweight neural networks using the SKD algorithm, KD-CNN and KD-MobileNet, to meet the demand that neural networks can be used on miniaturized devices. The simulation results demonstrate that the ClST outperforms advanced neural networks on all datasets. Moreover, both KD-CNN and KD-MobileNet obtain higher recognition accuracy with less network complexity, which is very beneficial for the deployment of AMR on miniaturized communication devices.

show abstract

Mixture GAN For Modulation Classification Resiliency Against Adversarial Attacks

Cited by 7 publications

References 12 publications

Effect of heavy metals on duckweed growth

Effect of heavy metals on duckweed growth

Towards Intelligent Security Mechanisms for Connected Things

ClST: A Convolutional Transformer Framework for Automatic Modulation Recognition by Knowledge Distillation

Contact Info

Product

Resources

About