Specific emitter identification (SEI) is a technique that identifies the unique emitter from its received signal by using the specific characteristics of an emitter. In this paper, we consider an SEI problem with unknown receiver distortion. Two groups of SEI schemes based on signal decomposition are proposed. In the proposed schemes, the received signal is pre-processed by either of the following decomposition, i.e., empirical mode decomposition (EMD), intrinsic time-scale decomposition (ITD), or variational mode decomposition (VMD). In the first group of the proposed schemes, the skewness and the kurtosis are extracted from the decomposed signal, which characterize the non-Gaussian features of the signal. The support vector machine (SVM) or the back-propagation (BP) neural network is applied to fuse the features extracted from the multiple distorted receivers respectively and then determine the unknown emitter. In the second group of the proposed schemes, an approach based on the long short term memory (LSTM) is proposed. The LSTM model learns the deep features rather than the specific non-Gaussian features from the pre-processed signal. In contrast to the first group, the features used to identify the unknown emitter are extracted directly from the pre-processed signal by the trained LSTM model. Simulation results show that the proposed multi-receiver cooperative schemes can achieve the diversity gain in the identification performance. Moreover, we evaluate the identification performance of the proposed schemes in various channels, including the Gaussian channel and the fading channel. Compared to the existing methods based on different time-frequency representations, the proposed schemes possess the merits of high identification accuracy and low complexity. The significance of this paper is that the receive diversity can be achieved by the proposed schemes by using multiple distorted receivers even without compensating the receiver distortion prior to the identification.

Specific emitter identification (SEI) methods via deep learning have shown significant progress in accuracy recently. However, these methods require a large amount of the labeled data. In this letter, contrastive learning is introduced to cope with the problem of the lack of the labeled data, which the network modules include the feature extractor, the projection head, and the classifier. We use a two-stage semi-supervised training scheme. In the first stage, the feature extractor extracts the features of the received signal, and the projection head deepens the network to reserve more information in the features. They are trained with the unlabeled samples via the self-supervised contrastive learning loss. In the second stage, the overall network is fine-tuned over a small amount of the labeled data using an alternative loss. This loss combines the original cross-entropy loss with the supervised contrastive learning loss. Numerical results on the FIT/CorteXlab dataset show that even with tens of the labeled samples, the accuracy of the proposed identifier attains around 90%, which outperforms the conventional supervised and semi-supervised identifiers.

This paper studies the joint secure transceiver design for the full-duplex integrated sensing and communication (ISAC) system, in which the base station performs the target tracking and communicates with the downlink and the uplink users by reusing the resources. Here, the target, referred to as Eve, is a potential eavesdropper with the intention of intercepting both the downlink and the uplink information. The security problem of the full-duplex ISAC system has not been studied well, where the sensing and communication signals suffer from the serious interference. In this paper, we jointly design the information beamformer, the radar waveform, the uplink communication receive filter, and the radar receive filter to achieve the downlink and uplink communication security and the target tracking. Specifically, both the Eve’s signal-to-interference-plus-noise ratio minimization and the secrecy rate maximization problems, subject to the sensing and the communication constraints, are formulated for the ISAC system from the perspectives of the quality of service and the secrecy rate. An iterative algorithm is proposed for solving the formulated problems. We prove that, under appropriate conditions, the proposed iterative algorithm converges to the Karush-Kuhn-Tucker optimal point of the original problem without the rank 1 constraint. We further extend the joint design to the case of the imperfect wiretap channel and the angle uncertainty. Numerical results show that our scheme remarkably outperforms the benchmark approaches. The performance is close to that of designing the secure communication and the sensing separately.

This letter explores the positive side of the adversarial attack for the security-aware semantic communication system. Specifically, a pair of matching pluggable modules is installed: one after the semantic transmitter and the other before the semantic receiver. The module at the transmitter uses a trainable adversarial residual network (ARN) to generate adversarial examples, while the module at the receiver employs another trainable ARN to remove the adversarial attack and the channel noise. To mitigate the threat of the semantic eavesdropping, the trainable ARNs are jointly optimized to minimize the weighted sum of the power of adversarial attacks, the mean squared error of the semantic communication, and the confidence of the eavesdropper correctly retrieving the private information. Numerical results show that our scheme can fool the eavesdropper while maintaining the high-quality semantic communication.

Specific emitter identification (SEI) is a technique to identify the unknown emitters by using the hardware impairment of the transmitter. In this paper, we consider the effect of the wireless channel on the SEI, which deteriorates the identification performance severely. Two identifiers are proposed to address the wireless channel effect from the model-based and the data-based perspectives, respectively. From the model-based perspective, the fingerprint extractor using the sparse Bayesian learning (SBL) is first proposed to jointly estimate the fingerprint parameters, the wireless channel, and the noise power in the multipath fading channels. Then, the classifier using the weighted Euclidean distance is designed to identify the unknown emitter. From the data-based perspective, the model-agnostic meta-learning (MAML) algorithm is adopted to meta-train the convolutional neural network (CNN) on the task collection, which is generated based on the transmitter distortion mechanism and the channel distribution. The trained CNN is fine-tuned on the unseen SEI task and then is used to identify the unknown emitter. Moreover, the Cramer-Rao lower bounds of the estimation of the fingerprint parameters are derived to evaluate the performance of the proposed fingerprint extractor. Numerical results show that the SBL identifier outperforms the MAML one in a small number of samples, while the MAML identifier outperforms the SBL one in a large number of samples. Both identifiers are robust to the wireless channel, obtain better identification performance, and require a small number of samples compared to the existing methods. Furthermore, the simulation results indicate that the mean squared error performance of the SBL fingerprint extractor is close to the performance lower bound.

For the modulation classification problems, the deep learning approaches can determine the unknown modulation formats in high confidence. However, it has been maliciously used by eavesdroppers. In this paper, we consider the wireless communication scenario, in which Alice intends to communicate with Bob confidentially in the threat of Eve, who tries to determine the unknown modulation formats of Alice using some deep learning approach. Recent advancements in adversarial machine learning have demonstrated that the deep learning techniques are vulnerable to crafted perturbations. To prevent Eve from classifying Alice’s modulation formats, Alice transmits the modulation signal with the well-designed adversarial perturbation. We first formulate an optimization problem to determine the optimized adversarial perturbation, in which the objective is to mislead the modulation classifier of Eve subject to the communication constraints, i.e., the power efficiency, the achievable rate, and the reliability. Then, the augmented Lagrangian method is adopted to solve the perturbation optimization problem, in which the implicit objective is evaluated using the Monte Carlo method, and the gradients of the implicit constraints are obtained using the Gaussian-based estimation algorithm. We further extend the perturbation design to the both cases of Alice having and not having the prior knowledge of Eve. Finally, the input-independent universal perturbation for the specific modulation type is proposed, which is deployed via a lookup table method. Numerical results show that the designed perturbation with 10% power of the modulated signal can attack Eve’s modulation classifier with the great success while ensuring both the achievable rate and the reliability close to the ideal case (say, no perturbation). Compared to the existing methods, the designed perturbation achieves the better attack performance and is robust to the filtering, the oversampling, and the time/frequency offset. Furthermore, this paper reveals that the structure type of Eve’s model has a large impact on the attack performance, and verifies that the adversarial perturbation can effectively attack the modulation classifiers that resort to the expert knowledge.