On the benefits of robust models in modulation recognition

Maroto, Javier; Bovet, Gérôme; Frossard, Pascal

doi:10.1117/12.2587156

Cited by 4 publications

(5 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In the case of modulation recognition, adversarial perturbations have also been shown to be effective and to require much less power than additive white gaussian noise (AWGN) to fool the network [8], [10], [11]. Adversarial perturbations in this case are constrained relatively to the signal power, using the signal-to-perturbation ratio (SPR) metric [8].…”

Section: Related Workmentioning

confidence: 99%

“…In modulation recognition, to have a realistic measure of how secure the model is against malicious attacks, the framework has to take into account that communication corruptions may be added between the attacker and the receiver side. This motivates us to use two different frameworks to measure robustness and security [11], which are shown in Figure 1. The robustness framework is similar to the traditional approach, adding the attack just before the model.…”

Section: Robustness Evaluation In Modulation Recognitionmentioning

confidence: 99%

“…We use the robustness framework described before with the VT CNN2 BF architecture [10] as our AMC model. For both the training and testing adversarial examples, we choose 20 dB for the SPR since it decreases significantly the performance of the model without compromising in perceptibility [11]. We use l ∞ PGA with 7 iterations of step size 0.36 (relative to the l ∞ ball radius) for the training adversarial examples, and PGA with 20 iterations of step size 0.125 for the testing adversarial examples.…”

Section: B Performance Analysismentioning

confidence: 99%

“…However, recent studies have highlighted security issues in DNNs models [6]- [11]. Specifically, they have been shown to be vulnerable to adversarial examples, which add a carefully crafted but almost imperceptible perturbation, namely adversarial perturbation, to a real data sample.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

SafeAMC: Adversarial training for robust modulation recognition models

Maroto,

Bovet,

Frossard

2021

Preprint

Self Cite

View full text Add to dashboard Cite

In communication systems, there are many tasks, like modulation recognition, which rely on Deep Neural Networks (DNNs) models. However, these models have been shown to be susceptible to adversarial perturbations, namely imperceptible additive noise crafted to induce misclassification. This raises questions about the security but also the general trust in model predictions. We propose to use adversarial training, which consists of fine-tuning the model with adversarial perturbations, to increase the robustness of automatic modulation recognition (AMC) models. We show that current state-of-the-art models benefit from adversarial training, which mitigates the robustness issues for some families of modulations. We use adversarial perturbations to visualize the features learned, and we found that in robust models the signal symbols are shifted towards the nearest classes in constellation space, like maximum likelihood methods. This confirms that robust models not only are more secure, but also more interpretable, building their decisions on signal statistics that are relevant to modulation recognition.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Robustness Evaluation In Modulation Recognitionmentioning

confidence: 99%

Section: B Performance Analysismentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

SafeAMC: Adversarial training for robust modulation recognition models

Maroto,

Bovet,

Frossard

2021

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…These recognition rates decrease severely in the real communication environment. According to the above-simulated datasets, a series of methods based on advanced signal features, such as time-frequency maps [ 12 ], constellation maps [ 13 ], waveform maps [ 14 ], instantaneous characteristics, and high-order cumulants [ 15 ], have been derived. However, these methods are essentially dependent on simulated datasets, and the recognition rates under noncooperative communication conditions cannot meet expectations.…”

Section: Introductionmentioning

confidence: 99%

Radio Frequency Database Construction and Modulation Recognition in Wireless Sensor Networks

Liu

Xiang

Yin

2022

Sensors

View full text Add to dashboard Cite

Current modulation recognition methods in wireless sensor networks rely too much on simulation datasets. Its practical application effect cannot reach the expected results. To address this issue, in this paper we collect a large amount of real-world wireless signal data based on the software radio device USRP 2920. We then propose a real radio frequency (RF) database architecture and preprocessing operators to manage real-world wireless signal data, conduct signal preprocessing, and export the dataset. Based on different feature datasets derived from the RF database, we propose a multidimensional feature hybrid network (MFHN), which is used to identify unknown signals by analyzing different kinds of signal features. Further, we improve MFHN and design a multifeatured joint migration network (MJMN) to identify small-sample targets. The experimental results show that the recognition rates for unknown target signals of the MFHN and MJMN are 82.7% and 93.2%, respectively. The proposed methods improve the recognition performance in the single node of wireless sensor networks in complex electromagnetic environments, which provides reference for subsequent decision fusion.

show abstract