Search

In cognitive radio (CR) networks, the automatic modulation classification (AMC) is considered as the significant role in smart wireless communications. Due to the high growth of deep learning in the modern days, neural network‐aided automated modulation categorization tasks have become highly demanded. Nevertheless, an enormous amount of attributes and the neural network's complexity make them complex to adopt in various scenarios. Moreover, the receiver systems are limited by the latency and less storage resources. Additionally, the detection system of the signal modulation is mostly hampered by overfitting issues and insufficient information. Therefore, a robust AMC model based on adaptive deep learning is implemented to determine the type of modulation used at the transmitter by observing the received signal. Initially, the necessary raw data for the suggested model is garnered from benchmark dataset. Also, the optimal features from the raw data are selected with the help of the fitness‐revised position updating in Kookaburra optimization (FPUKO) for minimizing the computation time and enhancing the accuracy rates in the classification process. Moreover, this optimal feature selection makes the modulation selection process quick and efficient. Finally, the optimal features are fed to a cascaded and attention‐based recurrent neural network (CA‐RNN) in the modulation classification, which is designed to classify the type of modulation used on the transmitter side. Various experiments are conducted for the designed framework by comparing it with the existing models to view its efficiency. [ABSTRACT FROM AUTHOR]

In this paper, we propose and experimentally demonstrate an artificial intelligence (AI)-enabled efficient modulation classification technique for underwater optical wireless communication (UOWC) systems. Specifically, time-domain waveform histograms are adopted as classification features, where three modulation formats including direct current biased optical orthogonal frequency division multiplexing (DCO-OFDM), asymmetrically clipped optical OFDM (ACO-OFDM) and pulse amplitude modulation (PAM) are considered. Moreover, AI algorithms such as decision trees (DT), k-nearest neighbors (k-NN), support vector machines (SVM) and convolutional neural networks (CNN) are utilized to realize efficient modulation classification based on the obtained waveform histogram features. Experimental results demonstrate that all the four algorithms can achieve accuracy surpassing 95% when the received signal-to-noise ratio (SNR) exceeds 6.3 dB. Furthermore, increasing the number of symbols in histograms enhances classification accuracy, whereas altering the number of histogram bins has minimal impact on classification accuracy. [ABSTRACT FROM AUTHOR]

Classifying modulated signals for non-cooperative underwater acoustic communication is challenging due to signal distortion caused by fading and multipath effects in the underwater acoustic channel. Our proposed method utilizes channel estimation parameters to measure and correct signal distortion, thereby enhancing the recognition performance of the received signal. Modulation classification experiments were conducted on a public dataset with various modulation schemes, as well as on the same dataset with simulated underwater acoustic channels. The results indicate that our method effectively mitigates the impact of the underwater acoustic channel on modulation signal classification, improves recognition accuracy, and is broadly applicable to a wide range of machine learning classifiers. Finally, we validated these findings using real underwater communication data. [ABSTRACT FROM AUTHOR]

Although modulation classification with deep learning has been widely explored, this is challenging when the training data is limited. In this paper, we meet this challenge by data augmentation based on a semi-supervised generative model, named semi-supervised variational auto-encoder GAN (SS-VAEGAN). The proposed model has two novel aspects: first, the ensemble prediction technique is presented to establish a semi-supervised framework, and thus few labeled data and abundant unlabeled data can be used simultaneously to train the generative model; second, a modified feature matching strategy is utilized to ensure the diversity of the generated samples and contribute to the final classification under limited labeled data. With the aim of the constellation diagram with density and the ResNet-18 network, we have worked out a complete solution of modulation classification with data augmentation using SS-VAEGAN. In the experiments, we investigate the classification accuracy and robustness of the proposed data augmentation method and then compare the proposed SS-VAEGAN with other deep generative models. It is shown that our method can achieve the classification accuracy of 0.968 in the extreme absence of training data. [ABSTRACT FROM AUTHOR]

In a Multiple-Input Multiple-Output Orthogonal Frequency Division Multiplexing (MIMO-OFDM) method, multiple antennas can be used on either the transmitter or receiver end to improve the system capacity, data throughput, and robustness. OFDM has been used as the modulation system that divides the data stream into multiple parallel low-rate subcarriers. MIMO enhances the system by utilizing spatial diversity and multiplexing abilities. Modulation classification in the MIMO-OFDM systems describes the process of recognizing the modulation scheme used by the communicated signals in a MIMO-OFDM communication system. This is a vital step in receiver design as it enables proper demodulation of the received signals. In this paper, an Enhanced Modulation Classification Approach using an Arithmetic Optimization Algorithm with Deep Learning (EMCA-AOADL) is developed for MIMO-OFDM systems. The goal of the presented EMCAAOADL technique is to detect and classify different types of modulation signals that exist in MIMO-OFDM systems. To accomplish this, the EMCA-AOADL technique performs a feature extraction process based on the Sevcik Fractal Dimension (SFD). For modulation classification, the EMCA-AOADL technique uses a Convolution Neural Network with Long Short-Term Memory (CNN-LSTM) approach. Finally, the hyperparameter values of the CNN-LSTM algorithm can be chosen by using AOA. To highlight the better recognition result of the EMCA-AOADL approach, a comprehensive range of simulations was performed. The simulation values illustrate the better results of the EMCA-AOADL algorithm.

In this paper, a novel fusion network for automatic modulation classification (AMC) is proposed in underwater acoustic communication, which consists of a Transformer and depth-wise convolution (DWC) network. Transformer breaks the limitation of sequential signal input and establishes the connection between different modulations in a parallel manner. Its attention mechanism can improve the modulation recognition ability by focusing on the key information. DWC is regularly inserted in the Transformer network to constitute a spatial-temporal structure, which can enhance the classification results at lower signal-to-noise ratios (SNRs). The proposed method can obtain more deep features of underwater acoustic signals. The experiment results achieve an average of 92.1% at -4 dB = SNR = 0 dB, which exceed other state-of-the-art neural networks. [ABSTRACT FROM AUTHOR]

Recently, deep learning (DL) based object detection methods have attracted significant attention for wideband multisignal detection, which has been viewed as an essential part in the field of cognitive radio spectrum sensing. However, the existing DL methods are difficult or very likely fail to detect discontinuous burst signals, not to mention the signals with wide, instantaneous, dynamic bandwidth, and multiple channels. To solve this problem, the present study proposes a scheme that combines the start–stop point signal features for wideband multi-signal detection, namely the Fast Spectrum-Size Self-Training network (FSSNet). Considering the horizontal rectangle form of a wideband signal in the time–frequency domain, we innovatively utilize the start–stop points of the two-dimensional (2D) Box to build the signal model. Specifically, We propose a fast Start–stop HeatMap where the proposed LPS-YXE simultaneously labels and divides the start–stop points positions in the X–Y axis of a single HeatMap. We attribute the method's success in discontinuous signal detection to the multidimensional space transformation of HeatMap, which is used to locate the start–stop points and extract features separated from the signal regions of start–stop points. Furthermore, FSSNet can realize the 2D Box estimation of the wideband signal by regressing only a single variable, and thus with satisfactory detection speed. Simulation results verify the effectiveness and superiority of the proposed start–stop based wideband signal detection scheme with practical received signals. All our models and code are available at https://github.com/jn-z/SSNet2. • By utilizing the start–stop points of signal's 2D boxes, a novel wideband signal detection network is proposed, which can be trained without special candidate anchors. • To the best of our knowledge, this is the first study to employ the start–stop points method for signal location in practical wideband signal communication scenario. • The proposed scheme seeks the global and local key feature information from start to endpoint, thus only regressing one variable from the signal regression branch, and reducing detection time significantly. [ABSTRACT FROM AUTHOR]

Modulation classification is a crucial technique to utilize the unconsumed spectrum in Cognitive Radio (CR) and Dynamic Spectrum Access (DSA) systems to meet the required traffic demands for future-generation cellular networks. This paper presents an end-to-end experimental setup as a generic methodology to implement various Transfer Learning (TL) models in an indoor environment. This allows us to learn the features from multiple modulation signals to train and test the model. The performance evaluation of proposed TL models such as Convolutional Neural Network-Random Forest (CNN-RF), and Convolutional Long Short Term Deep Neural Network (CLDNN) -Random Forest (CLDNN-RF) have been thoroughly discussed. The result shows that the proposed TL models yield more than 90% classification accuracy for various modulation types. A proposed framework for location-specific TL model selection based on the maximum classification accuracy has been investigated.

Classifying modulated signals for non-cooperative underwater acoustic communication is challenging due to signal distortion caused by fading and multipath effects in the underwater acoustic channel. Our proposed method utilizes channel estimation parameters to measure and correct signal distortion, thereby enhancing the recognition performance of the received signal. Modulation classification experiments were conducted on a public dataset with various modulation schemes, as well as on the same dataset with simulated underwater acoustic channels. The results indicate that our method effectively mitigates the impact of the underwater acoustic channel on modulation signal classification, improves recognition accuracy, and is broadly applicable to a wide range of machine learning classifiers. Finally, we validated these findings using real underwater communication data.

This paper introduces a new technique for automatic modulation classification (AMC) in Cognitive Radio (CR) networks. The method employs a straightforward classifier that utilizes high-order cumulant for training. It focuses on the statistical behavior of both analog modulation and digital schemes, which have received limited attention in previous works. The simulation results show that the proposed method performs well with different signal-to-noise ratios (SNRs) and channel conditions. The classifier's performance is superior to that of complex deep learning methods, making it suitable for deployment in CR networks' end units, especially in military and emergency service applications. The proposed method offers a cost-effective and high-quality solution for AMC that meets the strict demands of these critical applications. [ABSTRACT FROM AUTHOR]

After 70 years of intricate development, machine learning, represented by deep learning, is based on the multilevel structure of the human brain and the layer-by-layer analysis and processing mechanism of neuron connection and interaction information. The powerful parallel information processing ability of self-adaptation and self-learning has allowed for breakthroughs in many fields, among which the most representative is image recognition. Therefore, this paper proposed optimizing the RBF algorithm with machine learning (ML) to improve the recognition rate of spectrum sensing. The results showed that the average detection success rates of the RBF algorithm were 93.62%, 95.07%, 96.91%, 98.78% and 99.37% when the SNRs were − 8 dB, − 4 dB, 0 dB, 4 dB and 8 dB, respectively, and the other conditions were kept the same. The average detection success rates of the SVM/RBF algorithm were 97.65%, 99.63%, 99.76%, 99.91% and 99.88%, respectively. The average detection success rate of the SVM/RBF algorithm was significantly higher than that of the RBF algorithm. This indicates that analyzing the RBF neural network algorithm through ML can improve the success rate of spectrum sensing, which highlights a new direction for the application of ML and neural networks. [ABSTRACT FROM AUTHOR]

Wireless communication technologies are undergoing intensive study and are experiencing accelerated progress which leads to a large increase in the number of end-users. Because of this, the radio spectrum has become more crowded than ever. These previously mentioned aspects lead to the urgent need for more reliable and intelligent communication systems that can improve the spectrum efficiency. Specifically, modulation scheme recognition occupies a crucial position in the civil and military application, especially with the emergence of Software Defined Radio (SDR). The modulation recognition is an indispensable task while performing spectrum sensing in Cognitive Radio (CR). Spread spectrum (SS) techniques represent the foundation for the design of Cognitive Radio systems. In this work, we propose a new method of characterization of Spread spectrum modulations capable of providing relevant information for the process of recognition of this type of modulations. Using the proposed approach, results higher than 90% are obtained in the modulation classification process, thus bringing an advantage over the classical methods, whose performance is below 75%. [ABSTRACT FROM AUTHOR]

Modulation classification is a crucial technique to utilize the unconsumed spectrum in Cognitive Radio (CR) and Dynamic Spectrum Access (DSA) systems to meet the required traffic demands for future-generation cellular networks. This paper presents an end-to-end experimental setup as a generic methodology to implement various Transfer Learning (TL) models in an indoor environment. This allows us to learn the features from multiple modulation signals to train and test the model. The performance evaluation of proposed TL models such as Convolutional Neural Network - Random Forest (CNN-RF), and Convolutional Long Short Term Deep Neural Network (CLDNN) - Random Forest (CLDNN-RF) have been thoroughly discussed. The result shows that the proposed TL models yield more than 90% classification accuracy for various modulation types. A proposed framework for location-specific TL model selection based on the maximum classification accuracy has been investigated. [ABSTRACT FROM AUTHOR]

In recent years, neural network algorithms have demonstrated tremendous potential for modulation classification. Deep learning methods typically take raw signals or convert signals into time–frequency images as inputs to convolutional neural networks (CNNs) or recurrent neural networks (RNNs). However, with the advancement of graph neural networks (GNNs), a new approach has been introduced involving transforming time series data into graph structures. In this study, we propose a CNN-transformer graph neural network (CTGNet) for modulation classification, to uncover complex representations in signal data. First, we apply sliding window processing to the original signals, obtaining signal subsequences and reorganizing them into a signal subsequence matrix. Subsequently, we employ CTGNet, which adaptively maps the preprocessed signal matrices into graph structures, and utilize a graph neural network based on GraphSAGE and DMoNPool for classification. Extensive experiments demonstrated that our method outperformed advanced deep learning techniques, achieving the highest recognition accuracy. This underscores CTGNet's significant advantage in capturing key features in signal data and providing an effective solution for modulation classification tasks. [ABSTRACT FROM AUTHOR]

Automatic Modulation Classification (AMC) is a critical issue in electromagnetic spatial perception. Currently traditional recognition techniques are difficult to adapt to complex signal situations. Most existing modulation classification algorithms ignore the complementarity between different features and the importance of feature fusion. Based on this, we proposed a method for image feature fusion for AMC that fully uses the complementarity between different image features. The original signal is converted into an image by the Gramian Angular Field (GAF) method, and the GAF image is used as the input to the network, meanwhile the received signal is converted from the Inphase-Quadrature (I-Q) domain to the r- $\theta $ domain using the Accumulated Polar Feature conversion technique, and the original signal is feature coded from the r- $\theta $ domain and then converted into an image. The fused features of the two images are used as input to the neural network for model training to achieve automatic modulation classification of multiple types of signals. In the evaluation phase, the differences in the recognition effectiveness of the proposed method by different neural networks are discussed. Experiments show that the best performance is achieved using the Swin-Transformer network model, with a more than 90% recognition rate for the modulation method at signal-to-noise ratios greater than 4dB.

Automatic modulation classification (AMC) aims to automatically identify the modulation type of a detected signal in an intelligent wireless receiver, such as software-defined radio (SDR). Recently, deep learning-based methods such as convolutional neural networks have been applied to AMC, showing high-accuracy performance. However, the earlier studies do not consider various signal degradations that can possibly occur during the transmission and reception of wireless signals. Particularly, the signal reception can be often unstable, and the signal can be partially received due to the dynamic spectrum sensing or signal sensing in the intelligent wireless systems. The corrupted signal with missing samples considerably degrades the accuracy of modulation classification of the deep learning-based models, because it is very different from the training datasets. To address this issue, the preprocessing process of restoring the corrupted signal, called signal inpainting, is essential. Although it is significant for the modulation classification, no studies have been performed to investigate the effect of signal inpainting on AMC. To that end, this study proposes a generative adversarial network(GAN)-based signal inpainting method that fills in the missing samples in a wireless signal. The proposed inpainting method can restore the time-domain signal with up to 50% missing samples while maintaining the global structure of each modulation type. The correct recovery of the global structure enables the extraction of distinctive features that play a key role in the modulation classification. To investigate this effect of signal inpainting on AMC, we perform intensive experiments on the RadioML dataset that has been widely used in the AMC studies. We compare the accuracy performance of the two state-of-the-art AMC models with and without the proposed signal inpainting, respectively. Through the analysis of the results, we show that the proposed GAN-based inpainting method significantly improves the accuracy of AMC.

This study proposes robust convolutional neural network (CNN)-based automatic modulation classification (AMC) techniques. Traditional AMCs may be classified into two types: those that rely on ML (maximum likelihood-based AMCs) and those that rely on features. Numerous studies have been conducted on feature-based automatic modulation classification techniques. The current feature-based AMCs lack generalization capability and frequently target a small group of modulation techniques. The current paper develops three different CNN-based AMCs, each with a different classification layer (CL). The adopted classification layers are mean absolute error-based CL, a sum of squared errors-based CL, and crossentropy-based CL. The developed techniques can classify the received signals without feature extraction, where they can learn the features from the transmitted signals automatically during the offline training process, thus eliminating the necessity for feature extraction. A comparison study was done for the proposed CNN-based AMCs with three optimization algorithms at two signal-to-noise ratios. The proposed AMCs attain a true classification accuracy of up to 100% depending on the optimizer and loss function-base CL.

Modulation classification has been widely studied in recent years. However, few studies focus on the performance degradation in multipath fading channels, whose impact is non-negligible. In this paper, a convolutional neural network (CNN) employing cyclostationary (CS) feature, which maintain the essential characteristics in fading channels, is proposed for robust modulation classification. Our method can be implemented in two approaches, referred as CASE1 and CASE2. In CASE1, a single-structured CNN is designed for learning hybrid CS features to perform classification. And in CASE2, we present a CNN model based on a hierarchical structure to perform two-stage classification. Specifically, the coarse classification is performed by learning the second-order CS features with the first-level CNN. Next, another CNN can be selectively activated to learn from high-order CS features for fine classification within the subclass. In this way, our method uses CS features to provide favorable guidance for the learning process of CNN, thus improving the classification performance in fading channels. The experimental results demonstrate the advantages of the proposed method in terms of classification accuracy and computational complexity.

Summary: With the development of deep‐learning technology, how to prevent signal modulation from being correctly classified by deep‐learning‐based intruders becomes a challenging issue. Adversarial attack provides an ideal solution as deep‐learning models are proved to be vulnerable to intentionally designed perturbations. However, applying adversarial attacks to communication systems faces several practical problems such as shift‐invariant, imperceptibility, and bandwidth compatibility. To this end, a shift‐invariant universal adversarial attack approach is proposed in this work for misleading deep‐learning‐based modulation classifiers used by intruders. Specifically, this work first introduces a convolutional neural network (CNN)‐based UAP (universal adversarial perturbation) generation model that contains an finite impulse response (FIR) filter layer to control the bandwidth of the output perturbation. Besides, this work proposes a circular shift scheme that simulates the random signal cropping in the inference phase and thus ensures the shift‐invariant property of adversarial perturbations. In addition, this work designs a composite loss function that improves the imperceptibility of the adversarial perturbation in both time and frequency domains without decreasing the effectiveness of the adversarial attack. Experimental results demonstrate the effectiveness of the proposed approach, achieving about 50% accuracy drop on the target model when the perturbation‐to‐signal ratio (PSR) is −10 dB. Furthermore, extensive experiments are conducted to validate the shift‐invariant, imperceptibility, bandwidth compatibility, and transferability of the proposed approach for modulation classification tasks. [ABSTRACT FROM AUTHOR]

Visibility graph methods allow time series to mine non‐Euclidean spatial features of sequences by using graph neural network algorithms. Unlike the traditional fixed‐rule‐based univariate time series visibility graph methods, a symmetric adaptive visibility graph method is proposed using orthogonal signals, a method applicable to in‐phase and quadrature (I/Q) orthogonal signals for adaptive graph mapping for radio modulated signals in automatic modulation classification tasks. The method directly models the intra‐channel and inter‐channel graph relations of I/Q signals using two different types of convolutional kernels. It captures non‐Euclidean spatial feature information of I/Q signals using a graph neural network combining graph sampling aggregation and graph differentiable pooling as a feature extractor. Extensive experimental results on two benchmark datasets and a simulated dataset containing channel fading show that the proposed Quadrature Signal Symmetric Adaptive Visibility Graph (QSSAVG) method in this paper outperforms the benchmark method in terms of classification accuracy and is also more robust against channel fading and noise variations. [ABSTRACT FROM AUTHOR]

Convolutional Neural Networks (CNNs) are efficient tools for pattern recognition applications. They have found applications in wireless communication systems such as modulation classification from constellation diagrams. Unfortunately, noisy channels may render the constellation points deformed and scattered, which makes the classification a difficult task. This paper presents an efficient modulation classification algorithm based on CNNs. Constellation diagrams are generated for each modulation type and used for training and testing of the CNNs. The proposed work depends on the application of Radon Transform (RT) to generate more representative patterns for the constellation diagrams to be used for training and testing. The RT has a good ability to represent discrete points in the spatial domain as curved lines. Several pre-trained networks including AlexNet, VGG-16, and VGG-19 are used as classifiers for modulation type from the spatial-domain constellation diagrams or their RTs. Several simulation experiments are presented in this paper to compare different scenarios for modulation classification at different Signal-to-Noise Ratios (SNRs) and fading channel conditions. [ABSTRACT FROM AUTHOR]

Wireless communication technologies are undergoing intensive study and are experiencing accelerated progress which leads to a large increase in the number of end-users. Because of this, the radio spectrum has become more crowded than ever. These previously mentioned aspects lead to the urgent need for more reliable and intelligent communication systems that can improve the spectrum efficiency. Specifically, modulation scheme recognition occupies a crucial position in the civil and military application, especially with the emergence of Software Defined Radio (SDR). The modulation recognition is an indispensable task while performing spectrum sensing in Cognitive Radio (CR). Spread spectrum (SS) techniques represent the foundation for the design of Cognitive Radio systems. In this work, we propose a new method of characterization of Spread spectrum modulations capable of providing relevant information for the process of recognition of this type of modulations. Using the proposed approach, results higher than 90% are obtained in the modulation classification process, thus bringing an advantage over the classical methods, whose performance is below 75%.

The intelligent reflecting surface (IRS) is a cutting-edge technology for cost-effectively achieving future spectrum- and energy-efficient wireless communication. In particular, an IRS comprises many low-cost passive devices that can independently reflect the incident signal with a configurable phase shift to produce three-dimensional (3D) passive beamforming without transmitting Radio-Frequency (RF) chains. Thus, the IRS can be utilized to greatly improve wireless channel conditions and increase the dependability of communication systems. This article proposes a scheme for an IRS-equipped GEO satellite signal with proper channel modeling and system characterization. Gabor filter networks (GFNs) are jointly proposed for the extraction of distinct features and the classification of these features. Hybrid optimal functions are used to solve the estimated classification problem, and a simulation setup was designed along with proper channel modeling. The experimental results show that the proposed IRS-based methodology provides higher classification accuracy than the benchmark without the IRS methodology. [ABSTRACT FROM AUTHOR]

Intelligent methods have been applied to many fields for a long time. Recently, Visible Light Communication (VLC) systems widely include learning and classification models to improve their performances. The classification of L-Pulse Position Modulation (L-PPM) formats is crucial for VLC systems since the modulation order L is very effective for providing energy efficiency and increasing the transmission capacity. In this paper, therefore, it is reported for the first time, the classification of L-PPM schemes in VLC systems by using Decision Tree, K-nearest neighbor (KNN), Support Vector Machine, and a Direct Decision-based Linear Model technique. A novel feature extraction model is derived to be able to classify the type of L-PPM modulation schemes. A comparison has been given to observe the performance of classification schemes by taking into account the level of Signal to Noise Ratio and the transmission distance between receiver and transmitter. The 2-PPM scheme has the highest success rate against other schemes in terms of performance metrics at the overall distance. Additionally, the KNN method gives the best accuracy performance at the distance of 2.32 m and more, while it has been obtained the best accuracy of 97.85% by the Decision Tree Model at the distance of 2.20 m. [ABSTRACT FROM AUTHOR]

Integrating cognitive radio (CR) technique with wireless networks is an effective way to solve the increasingly crowded spectrum. Automatic modulation classification (AMC) plays an important role in CR. AMC significantly improves the intelligence of CR system by classifying the modulation type and signal parameters of received communication signals. AMC can provide more information for decision making of the CR system. In addition, AMC can help the CR system dynamically adjust the modulation type and coding rate of the communication signal to adapt to different channel qualities, and the AMC technique help eliminate the cost of broadcast modulation type and coding rate. Deep learning (DL) has recently emerged as one most popular method in AMC of communication signals. Despite their success, DL models have recently been shown vulnerable to adversarial attacks in pattern recognition and computer vision. Namely, they can be easily deceived if a small and carefully designed perturbation called an adversarial attack is imposed on the input, typically an image in pattern recognition. Owing to the very different nature of communication signals, it is interesting yet crucially important to study if adversarial perturbation could also fool AMC. In this paper, we make a first attempt to investigate how we can design a special adversarial attack on AMC. we start from the assumption of a linear binary classifier which is further extended to multi-way classifier. We consider the minimum power consumption that is different from existing adversarial perturbation but more reasonable in the context of AMC. We then develop a novel adversarial perturbation generation method that leads to high attack success to communication signals. Experimental results on real data show that the method is able to successfully spoof the 11-class modulation classification at a model with a minimum cost of about − 21 dB in automatic modulation classification task. The visualization results demonstrate that the adversarial perturbation manifests in the time domain as imperceptible undulations of the signal, and in the frequency domain as small noise outside the signal band. [ABSTRACT FROM AUTHOR]

The popular directions for automatic modulation classification algorithms investigate how to develop feature extraction methods for further signal classification. In this paper, we propose a mapping algorithm for a manually designed feature extraction method by using the reconstruction component of principal component analysis (PCA), which further extracts discrimination between signal features via a PCA reconstruction component. Two supervised neural network models are studied to achieve the limits of the learning matrix in modulation signal classification. Some experimental results show that different modulation schemes can be obviously classified using matrix mapping for feature extraction. Moreover, the modulation classification accuracy based on the mapping extraction feature, which has a lower SNR requirement for training, is slightly improved compared with some triradial deep learning methods. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Signal Processing is the property of Journal of Signal Processing and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

In this paper, we present a comprehensive survey and detailed comparison of techniques that have been applied to the problem of identifying the type of modulation contained within received wireless signals. Known as automatic modulation classification (AMC), the problem has been studied for many decades. AMC plays a significant role in both military and civilian scenarios and is the main step in smart receivers. With the development of software-defined radios and automatic communication systems, IoT technology and the spread of 5G technology, there has been exponential growth in the number of spectrum-using equipment making the issue of scarce spectrum resources more prominent. Although AMC techniques can be optimized from the classifier’s point of view, signal pre-processing also plays a critical role. Relevant data representation approaches include time-frequency analysis, cyclostationary transforms, and hybrid techniques. We provide a taxonomy of common approaches based on order and dimensionality along with an overall analysis of signal pre-processing algorithms for AMC. Furthermore, we reproduce the major existing schemes under uniform conditions, allowing an objective comparison among different methodologies. Finally, we create an open-source reproducible Python library to simulate these techniques, ensuring the usefulness for future research.

Many techniques have been developed for wireless signal recognition in many fifth generation (5G) enabled derivatives. Many harsh constraints, such as the large amount of model parameters and complex signal characteristics, drives intelligent recognition method in real-world settings. In this paper, we propose a generalizable, practical method for raw IQ signal recognition. Specifically, deep complex-valued convolutional neural network models, including a Complex-valued Visual Geometry Group (VGG) (CvVGG) model and a Complex-valued ResNet (CvRN) model, are proposed for handling raw signal IQ data. We examine the merit of complex-valued neural networks (CvNN) and validate their performance with experiments using two public datasets. With an SNR of 10dB, the proposed algorithm achieves a recognition accuracy of 96% on the RadioML2018.10a dataset. When performing drone recognition, CvNN can achieve a recognition accuracy of 99%. Our experimental results verify that deep complex-valued neural network models can achieve considerably improved accuracy with lower computation complexity and fewer model parameters than their real-valued counterparts.

Signal classification is a universal problem in adversarial wireless scenarios, especially when an eavesdropping radio receiver attempts to glean information about a target transmitter’s patterns, attributes, and contents over a wireless channel. In recent years, research surrounding the idea of Machine Learning (ML)-based signal classification has focused on modulation classification, with the downstream objective of demodulation. However, while the computer vision data domain has made significant progress in ensuring robust classification of images despite crafted perturbations, this success has not been translated to secure modulation classification. In this work, we perform the first-ever physical test of an eavesdropping ML-based modulation classifier radio, which we trained offline using a ensemble of i.i.d. models. Each model is trained with a weighted mixture of data perturbed by iterative, “least likely” white box attacks and non-attacked data. We then tested the ensemble online using coaxial-connected Software Defined Radios (SDRs). We conducted a case study comparing our results to the state-of-the-art computer vision approaches to investigate the presence of “label leaking”, model capacity sensitivity, understand the viability of parallel and sequential variations on perturbation training, and assess the effectiveness of iterative attack training. Our results show that perturbations can result in guessing-level classification performance from eavesdroppers, and that varying levels of robustness can be achieved against all presented attacks. These findings confirm that any receiver presents a new attack vector by utilizing ML techniques for classification tasks, and can be vulnerable to evasion attacks at little-to-no cost to transmitters. Consequently, we argue for the use of our training scheme in all ML-based classifying radios where security is a concern.

AbstractAutomatic modulation classification is essential in radar emitter identification. We propose a cascade classifier by combining a support vector machine (SVM) and convolutional neural network (CNN), considering that noise might be taken as radar signals. First, the SVM distinguishes noise signals by the main ridge slice feature of signals. Second, the complex envelope features of the predicted radar signals are extracted and placed into a designed CNN, where a modulation classification task is performed. Simulation results show that the SVM‐CNN can effectively distinguish radar signals from noise. The overall probability of successful recognition (PSR) of modulation is 98.52% at 20 dB and 82.27% at −2 dB with low computation costs. Furthermore, we found that the accuracy of intermediate frequency estimation significantly affects the PSR. This study shows the possibility of training a classifier using complex envelope features. What the proposed CNN has learned can be interpreted as an equivalent matched filter consisting of a series of small filters that can provide different responses determined by envelope features.

The Internet of Things (IoT) plays a significant role in building smart cities worldwide. Smart cities use IoT devices to collect and analyze data to provide better services and solutions. These IoT devices are heavily dependent on the network for communication. These new-age networks use artificial intelligence (AI) that plays a crucial role in reducing network roll-out and operation costs, improving entire system performance, enhancing customer services, and generating possibilities to embed a wide range of telecom services and applications. For IoT devices, it is essential to have a robust and trustable network for reliable communication among devices and service points. The signals sent between the devices or service points use modulation to send a password over a bandpass frequency range. Our study focuses on modulation classification performed using deep learning method(s), adaptive modulation classification (AMC), which has now become an integral part of a communication system. We propose a dependable modulation classifier explainer (DMCE) that focuses on the explainability of modulation classification. Our study demonstrates how we can visualize and understand a particular prediction made by seeing highlighted data points crucial for modulation class prediction. We also demonstrate a numeric explainability measurable metric (EMM) to interpret the prediction. In the end, we present a comparative analysis with existing state-of-the-art methods.

Automatic modulation recognition (AMR) has been wildly used in both military and civilian fields. Since the recognition of digital signal at low signal-to-noise (SNR) ratio is difficult and complex, in this paper, a clustering analysis algorithm is proposed for its recognition. Firstly, the digital signal constellation is extracted from the received waveform (digital signal + noise) by using the orthogonal decomposition and then, it is denoised by using an algorithm referred to as auto density-based spatial clustering technique in noise (ADBSCAN). The combination of density peak clustering (DPC) algorithm and improved K-means clustering is used to extract the constellation's graph features, the eigenvalues are input into cascade support vector machine (SVM) multi-classifiers, and the signal modulation mode is obtained. BPSK, QPSK, 8PSK, 16QAM and 32QAM five kinds of digital signals are trained and classified by our proposed method. Compared with the classical machine learning algorithm, the proposed algorithm has higher recognition accuracy at low SNR (less than 4dB), which confirmed that the proposed modulation recognition method is effective in noncooperation communication systems. [ABSTRACT FROM AUTHOR]

The recognition of modulation schemes in military and civilian applications is a major task for intelligent receiving systems. Various Automatic Modulation Classification (AMC) algorithms have been developed for this purpose in the literature. However, classification with low computational complexity as well as reasonable processing time is still a challenge. In this paper, a feature-based approach along with various classifiers is employed based on statistical features as well as higher-order moments and cumulants. An over-the-air (OTA) recorded dataset consisting of four analog and ten digital modulation schemes are used for testing the proposed method at 0–20 dB SNR. The overall accuracy for quadratic Support Vector Machine (SVM) is found to be as high as 98% at 10 dB. The comparison of the results with other AMC papers published in the literature indicates that the proposed method present higher accuracy, especially for realistic channel induced OTA dataset. [ABSTRACT FROM AUTHOR]

Modulation classification is a research hot-spot in the field of machine learning and the knowledge-based systems of wireless communication, involving the identification of different types of wireless signals. However, classification targets like radio signals usually involve long-sequence and large-scale data, and the classification of modulation types is affected by environmental noises, resulting in unsatisfactory performance. To overcome these challenges, we propose a novel Deep Hybrid Transformer Network (DH-TR) that makes full use of the intrinsic properties of multi-head self-attention and different neural modules, facilitating the identification of modulation types from global to local perspectives. In particular, DH-TR uses a convolution stem to extract local features inherent in In-phase and Quadrature (IQ) data, based on which the Gated Recurrent Unit (GRU) is used to capture the step-wise sequential patterns of these signals. Afterward, the self-attention based Transformer branch is employed to learn the global and long-term dependencies among the signal patches. With this innovative hybrid design, DH-TR performs well in processing sequential signal data and can better capture the complex relationships between signals. We validate DH-TR's effectiveness through extensive experiments on four benchmark datasets, demonstrating its superior performance in modulation classification with higher accuracy and robustness to noise compared to existing methods. • Study modulation classification, learning global to local signal representations. • Propose DH-TR for modulation classification, learning local and global features. • DH-TR outperforms state-of-the-art methods on four competitive modulation benchmarks. [ABSTRACT FROM AUTHOR]

Recurrent neural networks have been shown to outperform other architectures when processing temporally correlated data, such as from wireless communication signals. However, compared to other architectures, such as convolutional neural networks, recurrent neural networks can suffer from drastically longer training and evaluation times due to their inherent sample-by-sample data processing, while traditional usage of both of these architectures assumes a fixed observation interval during both training and testing, the sample-by-sample processing capabilities of recurrent neural networks opens the door for alternative approaches. Rather than assuming that the testing and observation intervals are equivalent, the observation intervals can be "decoupled" or set independently. This can potentially reduce training times and will allow for trained networks to be adapted to different applications without retraining. This work illustrates the benefits and considerations needed when "decoupling" these observation intervals for spectrum sensing applications, using modulation classification as the example use case. The sample-by-sample processing of RNNs also allows for the relaxation of the typical requirement of a fixed time duration of the signals of interest. Allowing for variable observation intervals is important in real-time applications like cognitive radio where decisions need to be made as quickly and accurately as possible as well as in applications like electronic warfare in which the sequence length of the signal of interest may be unknown. This work examines a real-time post-processing method called "just enough" decision making that allows for variable observation intervals. In particular, this work shows that, intuitively, this method can be leveraged to process less data (i.e., shorter observation intervals) for simpler inputs (less complicated signal types or channel conditions). Less intuitively, this works shows that the "decoupling" is dependent on appropriate training to avoid bias and ensure generalization. [ABSTRACT FROM AUTHOR]

Automatic modulation classification finds its application in many military and civil areas. It is an integral part of cognitive radio and software defined radio technologies. In this paper, LabVIEW based Field Programmable Gate Array (FPGA) implementation of modulation classification algorithm is proposed. Any modulation scheme among BPSK, QPSK, 8PSK, 8QAM, 16QAM and 4ASK is classified by alteration of oversampling factor and further error minimization between extracted constellation and ideal constellation of considered modulation schemes. Study results reveal that the developed method detects the above mentioned modulation schemes reliably above 12 dB SNR in Additive White Gaussian Noise (AWGN) channel. Comparative analysis of the proposed method with existing methods based on higher order statistics, mel-frequency ceptral coefficients and naive based modulation classification shows an overall improvement in classification accuracy. [ABSTRACT FROM AUTHOR]

Deep learning is increasingly being used for many tasks in wireless communications, such as modulation classification. However, it has been shown to be vulnerable to adversarial attacks, which introduce specially crafted imperceptible perturbations, inducing models to make mistakes. This letter proposes an input-agnostic adversarial attack technique that is based on generative adversarial networks (GANs) and multi-task loss. Our results show that our technique reduces the accuracy of a modulation classifier more than a jamming attack and other adversarial attack techniques. Furthermore, it generates adversarial samples at least 335 times faster than the other techniques evaluated, which raises serious concerns about using deep learning-based modulation classifiers. [ABSTRACT FROM AUTHOR]

This paper presents channel-aware adversarial attacks against deep learning-based wireless signal classifiers. There is a transmitter that transmits signals with different modulation types. A deep neural network is used at each receiver to classify its over-the-air received signals to modulation types. In the meantime, an adversary transmits an adversarial perturbation (subject to a power budget) to fool receivers into making errors in classifying signals that are received as superpositions of transmitted signals and adversarial perturbations. First, these evasion attacks are shown to fail when channels are not considered in designing adversarial perturbations. Then, realistic attacks are presented by considering channel effects from the adversary to each receiver. After showing that a channel-aware attack is selective (i.e., it affects only the receiver whose channel is considered in the perturbation design), a broadcast adversarial attack is presented by crafting a common adversarial perturbation to simultaneously fool classifiers at different receivers. The major vulnerability of modulation classifiers to over-the-air adversarial attacks is shown by accounting for different levels of information available about the channel, the transmitter input, and the classifier model. Finally, a certified defense based on randomized smoothing that augments training data with noise is introduced to make the modulation classifier robust to adversarial perturbations. [ABSTRACT FROM AUTHOR]

Abstract This paper proposes a new approach for finding the conditionally optimal solution (the classifier with minimum error probability) for the classification problem where the observations are from the multivariate normal distribution. The optimal Bayes classifier does not exist when the covariance matrix is unknown for this problem. However, this paper proposes a classifier based on the constant false alarm rate (CFAR) and invariance property. The proposed classifier is optimal conditionally as it has the minimum error probability in a subset of solutions. This approach has an analogy to hypothesis testing problems where uniformly most powerful invariant (UMPI) and uniformly most powerful unbiased (UMPU) detectors are used instead of the non-existing optimal UMP detector. Furthermore, this paper investigates using the proposed classifier for modulation classification as an application in signal processing.

Operators of modern mobile networks are faced with significant challenges in providing the requested level of service to an ever increasing number of user entities. Advanced machine learning techniques based on deep architectures and appropriate learning methods are recognized as promising ways of tackling the said challenges in many aspects of mobile networks, such as mobile data and mobility analysis, network control, network security and signal processing. Having firstly presented the background of deep learning and related technologies, the paper goes on to present the architectures used for deployment of deep learning in mobile networks. The paper continues with an overview of applications and services related to the new generation of mobile networks that employ deep learning methods. Finally, the paper presents practical use case of modulation classification as implementation of deep learning in an application essential for modern spectrum management. We complete this work by pinpointing future directions for research.

With the proliferation of frequency-using devices and the advent of the era of big data, spectrum management and control are faced with challenges of effectiveness and accuracy. Modulation classification technology is the foundation and key part of spectrum management and control. Therefore, the effectiveness of modulation classification technology in big data scenario is very important. In this paper, we consider not only the validity of the classification model under the background of big data, but also the dynamics of noise in the complex electromagnetic environment. So we construct a big dataset containing different signals under different MSNR, and use the big data to drive the deep learning model, and finally get the classification result. The proposed method can realize modulation classification only by training one model, which avoids the redundancy of model training in previous algorithms. The simulation results demonstrate the effectiveness and reliability of the proposed method.