In this paper, the problem of transmit sequence design for a dual-function radar-communication (DFRC) system equipped with one-bit digital-to-analog converters (DACs) is investigated. More specifically, a nonconvex problem is formulated by minimizing the symbol mean-square error, while ensuring the target localization performance for radar. A direct one-bit sequence design tackled by an alternating minimization (AM) approach, which involves a simple unconstrained quadratic sub-problem with closed-form solution and a quadratically constrained nonconvex sub-problem tackled by the alternating direction method of multipliers (ADMM) algorithm, is developed. The solutions of primal variables under the ADMM framework are provided and the convergence is discussed. Additionally, an indirect but computationally more efficient design, which is realized by transmit beamforming based on an accelerated primal gradient (APG) method, is presented. For better understanding of this design, both the Cramér-Rao Bound (CRB) and symbol error probability of the resulting beamformer with one-bit quantization using the Bussgang theorem are analyzed. Numerical simulations are provided to demonstrate the effectiveness of the proposed schemes.

This paper considers the problem of co-existence of collocated multiple-input multiple-output (MIMO) radar and downlink multiple-input single-output (MISO) communication systems, which share the same frequency band. As the interference caused by the communication systems would degrade the accuracy of radar target localization, we formulate the co-design of radar waveform and communication transmit weights by minimizing the Cramér-Rao Bound (CRB) of direction-of-arrival (DOA) estimation, subject to a set of constraints accounting for the worst-case signal-to-noise-plus-interference ratio (SINR), similarity and energy. However, the objective is nonconvex and all variables are coupled together in the SINR constraints, the formulated problem is thus NP-hard. Towards that end, a decentralized block successive upper-bound minimization (DBSUM) method based on the decomposition theory is developed. More specifically, at each iteration of this algorithm, the radar waveform is obtained with the aid of the alternating direction method of multipliers (ADMM) algorithm, and communication weights are obtained by exploiting the semidefinite programs (SDP). It is also proved that the proposed SDP-based method gives a rank-one solution. Numerical simulations are conducted to evaluate the effectiveness of the proposed algorithm.

Specific emitter identification (SEI) based on radio frequency fingerprint (RFF) characteristics can be used to identify different transmitters, and the deep learning (DL)-based SEI methods have achieved good performance with sufficient samples. However, these methods are difficult to identify emitters when the labeled training samples are limited. Hence, we propose a few-shot SEI (FS-SEI) using asymmetric masked auto-encoder (AMAE) to solve the few-shot problem. Specifically, we use the sufficiently unlabeled training samples which refer as the source domain to drive the training process of AMAE to obtain an RFF extractor with good feature extraction performance on the source domain, and then the pre-trained RFF extractor together with a classifier is fine-tuned using limited labeled training samples which refer as the target domain. Simulation results show that the proposed AMAE-based FS-SEI method achieves state-of-the-art identification performance compared to other supervised and unsupervised methods on the LoRa dataset with 30 categories and WiFi dataset with 16 categories. The codes can be downloaded from GitHub: https://github.com/YZS666/A-Method-for-Solving-the-FS-SEI-Problem.

Deep learning (DL)-based specific emitter identification (SEI) is a potential physical layer authentication technique for Industrial Internet-of-Things (IIoT) Security, which detects the individual emitter according to its unique signal features resulting from transmitter hardware impairments. The success of DL-based SEI often depends on sufficient training samples and the integrity of samples’ labels. The extensive deployment of wireless devices generates a huge amount of signals, but signals labeling is quite difficult and expensive with the high demand for expertise. In this article, we present an SEI method based on dual consistency regularization (DCR), which enables feature extraction and identification using a few labeled samples and a large number of unlabeled samples. With the help of pseudo labeling, we leverage consistency between the predicted class distribution of weakly augmented unlabeled training samples and that of strongly augmented training unlabeled samples, and consistency between semantic feature distribution of labeled samples and that of pseudo-labeled samples, which takes the unlabeled samples into account to model parameter tuning for a more accurate emitter identification. Extensive numerical results demonstrate that compared with well-known semi-supervised learning-based SEI methods, our method obtains 99.77% identification accuracy on a WiFi data set and 90.10% identification accuracy on an automatic dependent surveillance-broadcast (ADS-B) data set when only 10% of training samples are labeled, and improves the identification accuracy on the WiFi data set and the ADS-B data set by more than 19.07% and 5.30%, respectively. Our codes are available at https://github.com/lovelymimola/DCR-Based-SemiSEI .

Radio spectrum prediction is an important task for dynamic spectrum management and spectrum congestion mitigation. However, due to the complexity and spatiotemporal variability of spectrum data, spectrum prediction has always been a challenging problem. To improve prediction accuracy, this paper proposes a multi-band spectrum prediction method based on attention graph convolutional recurrent neural networks (A-GCRNN), which applies temporal correlation and frequency band correlation to spectrum prediction tasks. This method represents the spectrum data of multiple bands as a graph, with each band corresponding to a node in the graph, and utilizes graph convolutional networks (GCN) to learn the correlations between different bands. Then, a gated recurrent unit (GRU) network is employed to capture temporal correlations for each band, thereby fusing the acquired feature information to predict the future spectrum of each frequency band. In addition, an attention network is used to weight the output hidden states and further enhance the model's convergence rate. We conduct experiments on a real-world spectrum dataset and compare our method with other spectrum prediction methods. The results demonstrate that our approach achieves better performance in multi-band spectrum prediction tasks. Our pytorch implementation of the A-GCRNN is available at https://github.com/TLTLHILL/A-GCRNN-for-spectrum-prediction .

This paper deals with the waveform design to ensure the spectrum co-existence between multiple-input multiple-output (MIMO) radar and communication systems. The transmit waveform energy of MIMO radar in the prescribed spatial-frequency domain is minimized to reduce the mutual interference induced by the overlaid communication services. Additionally, the integrated sidelobes level (ISL) and peak-to-average power ratio (PAPR) constraints are forced on the probing waveform. To tackle the resulting NP-hard problem, we first introduce several auxiliary variables to convert the fourth-order ISL constraint and the PAPR constraint into several quadratic constraints. Then, the recast problem is formulated into a new alternating direction method of multipliers (ADMM) form, with small-sized tractable subproblems solved sufficiently. Numerical simulations verify that the proposed algorithm outperforms another competitive method in terms of the convergence performance and also demonstrate that the designed waveform can achieve spatial-frequency nulling under the ISL and PAPR constraints.