Digital Modulation Classification Research Articles

Deep multilevel architecture for automatic modulation classification

In the realm of wireless communication, deep learning has exhibited promising outcomes across various tasks, including automatic modulation classification (AMC), channel estimation, and signal detection, yielding substantial enhancements in performance and efficiency. AMC plays a pivotal role in identifying the modulation scheme of a received signal without prior knowledge of the transmitter. However, existing AMC techniques attempt to classify both analog and digital modulations together, disregarding their distinct characteristics. In this work, we propose a multilevel approach for AMC, wherein the received signals are first classified into their parent class, i.e., analog or digital modulation, and subsequently further classified into their respective modulation type. For implementing multilevel approach, we propose three distinct deep learning models: analog digital modulation classifier (AD-MC), analog modulation classifier (An-MC), and digital modulation classifier (Dig-MC). Performance analysis is conducted for the individual models as well as for the multilevel model that integrates all three models. The multilevel model demonstrates accurate signal classification at Signal-to-Noise Ratios (SNRs) of 0 dB or higher, achieving a remarkable classification accuracy of over 90%. Furthermore, the proposed multilevel model surpasses state-of-the-art AMC methods in accuracy while maintaining lower complexity.

Robust Automatic Modulation Classification Using Convolutional Deep Neural Network Based on Scalogram Information

This study proposed a two-stage method, which combines a convolutional neural network (CNN) with the continuous wavelet transform (CWT) for multiclass modulation classification. The modulation signals’ time-frequency information was first extracted using CWT as a data source. The convolutional neural network was fed input from 2D pictures. The second step included feeding the proposed algorithm the 2D time-frequency information it had obtained in order to classify the different kinds of modulations. Six different types of modulations, including amplitude-shift keying (ASK), phase-shift keying (PSK), frequency-shift keying (FSK), quadrature amplitude-shift keying (QASK), quadrature phase-shift keying (QPSK), and quadrature frequency-shift keying (QFSK), are automatically recognized using a new digital modulation classification model between 0 and 25 dB SNRs. Modulation types are used in satellite communication, underwater communication, and military communication. In comparison with earlier research, the recommended convolutional neural network learning model performs better in the presence of varying noise levels.

Digital Modulation Classification Based On Chicken Swarm Optimization and J48 Algorithm

Automatic Modulation Recognition (AMR) has a significant impact in the military as well as civil applications. Recognizing the modulation of the received signal has been considered as an intermediate step between the detection and demodulation of the signal. Which is why, in many military and communication systems, the AMR is considered as part of the system. Presently, due to increasing digital modulations in military and civil applications. Digital modulation recognition is especially important. Usually for the AMR a small number of the received signal features are obtained and utilized. The choice of the suitable feature plays an important part in the increase of AMR efficiency. The presented paper indicates hybrid intelligent system for the recognitions of digital signal types, consisting of 3 major modules: classifier module, feature extraction module and J48 Classifier that was used for the first time in our research in the field of classification of modulated signals and optimization module by Chicken Swarm Optimization (CSO). To get better results of the system suggested optimization the features to discard weak or irrelevant features in the system and keep only strong relevant features Chicken Swarm Optimization. The results of simulation confirm the high accuracy of recognition that is related to the suggested system even at low SNR.

Hierarchical Digital Modulation Classification Using Cascaded Convolutional Neural Network

Automatic modulation classification (AMC) aims to identify the modulation format of the received signals corrupted by the noise, which plays a major role in radio monitoring. In this paper, we propose a novel cascaded convolutional neural network (CasCNN)-based hierarchical digital modulation classification scheme, where M-ary phase shift keying (PSK) and M-ary quadrature amplitude modulation (QAM) modulation formats are considered to be classified. In CasCNN, two-block convolutional neural networks are cascaded. The first block network is utilized to classify the different classes of modulation formats, namely PSK and QAM. The second block is designed to identify the indexes of the modulations in the same PSK or QAM class. Moreover, it is noted that the gird constellation diagram extracted from the received signal is utilized as the inputs to the CasCNN. Extensive simulations demonstrate that CasCNN yields performance gain and performs stronger robustness to frequency offset compared with other recent methods. Specifically, CasCNN achieves 90% classification accuracy at 4 dB signal-to-noise ratio when the symbol length is set as 256.

Digital Signal Modulation Classification using Data Conversion Method based on Convolutional Neural Network

Automated modulation classification (AMC) plays a very important part in cognitive electronic warfare. Deep learning (DL) is a new machine learning (ML) method, which has been successfully implemented in many application fields. This paper proposes a new data conversion algorithm in order to gain a better classification accuracy of target signal modulation. Then a Convolutional Neural Network (CNN) architecture is developed, and its performance is proved to be better than the traditional data conversion method. In addition, the impacts of representation on classification performance are also analysed. Experimental results demonstrate the significant performance advantage of the proposed data conversion method using DL-based approach for modulation classification.

Digital Modulation Classification Based On BAT Swarm Optimization and Random Forest

The applications of digitally modulated signals are still in progress and expansion. Automatic Modulation Identification (AMI) is important to classify the digitally modulated signals ..To get better results of the system suggested optimization the features to discard weak or irrelevant features in the system and keep only strong relevant features .In this work, present hybrid intelligent system for the recognition related to the digitally modulated signals where used . The proposed (AMI) had been built to classify ten most popular schemes of digitally modulated signals, namely (2ASK, , 2PSK, 4PSK, 8PSK, 8QAM,16QAM ,32QAM, 64 QAM, 128QAM, and 256QAM), with the signal to noise ratio ranging from (-2 to 13) dB. High-order cumulants (HOCs) as well as high-order moments (HOMs) were utilized. .In this thesis used , Bat Swarm Optimization (BA).The Random Forest ( RF) classifier was introduced for the first time in this work. Simulation results of the System proposed , under additive white Gaussian noise channel, show that .While algorithm ( BA Swarm Optimization ) for the modulated signals we obtained a classification accuracy of around 92% for the SNR between (-2....12 ) dB.

Automatic determination of digital modulation types with different noises using Convolutional Neural Network based on time–frequency information

In this study, a novel digital modulation classification model has been proposed for automatically recognizing six different modulation types including amplitude shift keying (ASK), frequency shift keying (FSK), phase-shift keying (PSK), quadrate amplitude shift keying (QASK), quadrate frequency shift keying (QFSK), and quadrate phase-shift keying (QPSK). The determination of modulation type is significant in military communication, satellite communication systems, and submarine communication. To classify the modulation types, we have proposed a two-stage hybrid method combining short-time Fourier transform (STFT) and convolutional neural network (CNN). In the first stage, as the data source, the time–frequency information from these modulation signals have been extracted with STFT. This information has been obtained as 2D images to feed the input of the CNN deep learning method. In the second stage, the obtained 2D time–frequency information has been given to the input of the CNN algorithm to classify the modulation types. In this work, noises at various SNR values from 0 dB to 25 dB were created and added to the modulated signals. Even in the presence of noise, the proposed hybrid deep learning model achieved excellent results in the noised-modulation signals.

Blind digital modulation classification for STBC‐OFDM system in presence of CFO and channels estimation errors

Here, the authors propose a robust blind digital modulation classification (BDMC) algorithm for space time block coding (STBC)-based MIMO-OFDM system in the presence of carrier frequency offset (CFO) and channel estimation errors. Previous papers published on the topic of modulation identification were limited to single-carrier systems operating over frequency-flat channels. The problem of joint channel and CFO estimation in conjunction with blind digital modulation classification for STBC-OFDM in frequency selective channel and in the presence of the impulsive noise has not been addressed before to the best of their knowledge. To cope with performance degradation of BDMC due to CFO and channels errors, they propose joint semi-blind CFO and channels estimation methods. Higher order statistics (HOS), used for feature extraction, are combined with pattern recognition methods to solve the modulation identification problem. The main contribution of their work is the development of estimators, the study of their impacts on the blind classification capability, and the use of simulations to demonstrate the superior performance of the proposed algorithms.

Automatic digital modulation classification using extreme learning machine with local binary pattern histogram features

Discrimination of the Local Binary Pattern (LBP) in the classification of different digital modulation types was investigated in this study. It has been shown that LBP can be used as a feature extraction method for AMC schemes. A new AMC scheme is proposed using Extreme Learning Machine (ELM) as a classifier, which has a faster learning process and better generalization performance than conventional machine learning methods. The study also investigated the stability of the proposed AMC scheme, which is affected by variation in the values of the roll-off factor, frequency and phase offset that can affect the stability and performance of the system. Through simulation, a classification accuracy of over 95% was achieved at low SNR levels such as −2 dB. It was also shown that the proposed AMC scheme is more successful under similar conditions when making comparisons to other studies.

Automatic Digital Modulation Classification Based on Curriculum Learning

Neural network shows great potential in modulation classification because of its excellent accuracy and achievability but overfitting and memorizing data noise often happen in previous researches on automatic digital modulation classifier. To solve this problem, we utilize two neural networks, namely MentorNet and StudentNet, to construct an automatic modulation classifier, which possesses great performance on the test set with −18–20 dB signal-to-noise ratio (SNR). The MentorNet supervises the training of StudentNet according to curriculum learning, and deals with the overfitting problem in StudentNet. The proposed classifier is verified in several test sets containing additive white Gaussian noise (AWGN), Rayleigh fading, carrier frequency offset and phase offset. Experimental results reveal that the overall accuracy of this classifier for common eleven modulation types was up to 99.3% while the inter-class accuracy could be up to 100%, which was much higher than many other classifiers. Besides, in the presence of interferences, the overall accuracy of this novel classifier still could reach 90% at 10 dB SNR indicting its excellent robustness, which makes it suitable for applications like military electronic warfare.

Digital Signal Modulation Classification With Data Augmentation Using Generative Adversarial Nets in Cognitive Radio Networks

Automated modulation classification plays a very important part in cognitive radio networks. Deep learning is also a powerful tool that we could not overlook its potential in addressing signal modulation recognition problem. In our last work, we propose a new data conversion algorithm in order to gain a better classification accuracy of communication signal modulation, but we still believe that the convolution neural network (CNN) can work better. However, its application to signal modulation recognition is often hampered by insufficient data and overfitting. Here, we propose a smart approach to programmatic data augmentation method by using the auxiliary classifier generative adversarial networks (ACGANs). The famous CNN model, AlexNet, has been utilized to be the classifier and ACGAN to be the generator, which will enlarge our data set. In order to alleviate the common issues in the traditional generative adversarial nets training, such as discriminator overfitting, generator disconverge, and mode collapse, we apply several training tricks in our training. With the result on original data set as our baseline, we will evaluate our result on enlarged data set to validate the ACGAN’s performance. The result shows that we can gain 0.1~6% increase in the classification accuracy in the ACGAN-based data set.

Parameter estimation for blind classification of digital modulations

In this study, the performance of the likelihood-based digital modulation classification is explored with the blind estimation of the unknown parameters. Considering the practical implementation aspects, the quasi hybrid likelihood ratio test (QHLRT) is examined with the symbol rate, the signal gain, the noise power and the phase offset as the unknown parameters. In a blind scenario, new algorithms are proposed for the estimation of the unknown parameters with special focus on the improvement of classification performance at a low signal to noise ratio scenario. The performance bounds of the proposed estimators are established by the Cramer–Rao lower bound. The proposed method is compared with several existing algorithms to analyse the improvements achieved in the slow fading scenario. With the estimates of the unknown parameters, the performance of the QHLRT classifier is presented with reference to the theoretical upper bound. Finally, the QHLRT based method with the proposed parameter estimators is compared with the existing LB as well as certain feature based algorithms to highlight the improvements achieved.

Distinguishing CPFSK from QAM and PSK Modulations

Digital modulation classification is important for many civilian as well as military applications. In this paper, we propose a simple and robust feature to distinguish continuous-phase FSK from QAM and PSK modulations. The feature is based on product of two consecutive signal values and on time averaging of imaginary part of the product. Conditional probability density functions of the feature given modulation type are determined. In order to overcome the complexity of calculating probability density functions, central limit theorem for strictly stationary m-dependent sequences is used to obtain Gaussian approximations. After calculating probability density functions, thresholds are determined based on minimization of total probability of misclassification. Since threshold-based results are valid for special cases requiring knowledge of some parameters, we resort to usage of support vector machines for classification, which require little training and no a priori information except for carrier frequency. Following that joint effects on the performance of carrier offset, fast fading, and non-synchronized sampling are studied in the presence of additive white Gaussian noise. For comparison purposes, rectangular pulse shape is used. To prove practical usefulness, not only the performance is analyzed for root-raised cosine pulses but also for quite less oversampling of symbols than what is found in other approaches. In the course of doing that, the performance is compared with wavelet-based feature that uses support vector machines for modulation separation.

A new cumulant estimator in multipath fading channels for digital modulation classification

In this study, the authors consider the problem of digital modulation classification in a multipath channel using the cumulant-based method. A new estimator is proposed to estimate the fourth-order cumulant of transmitted symbol in a multipath channel with the help of the fourth-order cross-cumulants of the received symbols. The proposed estimator does not need the estimation of the multipath channel coefficients and is shown to be asymptotically unbiased. The channel order is an important factor to affect the estimator. The impact of the channel order mismatch on the estimator is analysed and it is proved that the estimator keeps unbiased in the case of the overestimation of the channel order. The simulated experiments are made to verify the cumulant-based classifier using the new estimator and the results show that it attains better performance for digital modulation classification than the two existing cumulant-based classifiers in a multipath channel.

Digital Modulation Classifier with Rejection Ability via Greedy Convexhull Learning and Alternative Convexhull Shrinkage in Feature Space

Automatic modulation classification is a key technique in wireless communication systems. In practical scenario, the modulation format of a received signal may be new or unknown to classifiers. However, most of existing classifiers compulsively classify the received signal as one of the candidate modulation formats. The rejection ability of a classifier to unknown modulation formats is important in applications. In this paper, a fourth-order cumulant-based classifier with rejection ability is proposed to classify digital modulation formats under additive white Gaussian noise channel. The two fourth-order cumulants are used as the feature vector. The classification with rejection ability boils down to the segmentation problem of the two-dimensional feature space with rejection region. A two-stage optimization is proposed to attain the suboptimal solution of the problem. The greedy convexhull learning algorithm is used to determine the primary decision regions of all the candidate modulation formats from training sets. The primary decision regions have rejection ability but are not always mutually separate. The alternative convexhull shrinkage is presented to separate the primary decision regions at small loss in probability of correct classification (PCC). The proposed classifier is compared with the cumulant-based hierarchical classifier and the K-S classifiers. The results show that besides desired rejection ability it is competitive with these classifiers in PCC.

Likelihood-Based Modulation Classification for Multiple-Antenna Receiver

Likelihood-based algorithms for the classification of linear digital modulations are systematically investigated for a multiple receive antennas configuration. Existing modulation classification (MC) algorithms are first extended to the case of multiple receive antennas and then a critical problem is identified that the overall performance of the multiple antenna systems is dominated by the worst channel estimate of a particular antenna. To address the performance degradation issue, we propose a new MC algorithm by optimally combining the log likelihood functions (LLFs). Furthermore, to analyze the upper-bound performance of the existing and the proposed MC algorithms, the exact Cramer-Rao Lower Bound (CRLB) expressions of non-data-aided joint estimates of amplitude, phase, and noise variance are derived for general rectangular quadrature amplitude modulation (QAM). Numerical results demonstrate the accuracy of the CRLB expressions and verify that the results reported in the literature for quadrature phase-shift keying (QPSK) and 16-QAM are special cases of our derived expressions. Also, it is demonstrated that the probability of correct classification of the new algorithm approaches the theoretical bounds and a substantial performance improvement is achieved compared to the existing MC algorithm.

A algorithm of fast digital phase modulation signal recognition

Classifications of digital modulation, the signal characteristics of the phase of the cycle will cause the phase wrapped and reduce overlapping signals in the received RF digital modulation automatically correct classification probability. According to the principle of minimum mean square, a phase modulation and phase modulation signal classification features as well as 2PSK and MPSK classification. The experiments illustrate the robust of the feature-based approach for automatic digital modulation classification and a higher recognition probability suitable to various pulse shaping filters, roll-off factors and symbol rates. DOI: http://dx.doi.org/10.11591/telkomnika.v10i8.1666

Automatic modulation classification of digital modulations in presence of HF noise

Designing an automatic modulation classifier (AMC) for high frequency (HF) band is a research challenge. This is due to the recent observation that noise distribution in HF band is changing over time. Existing AMCs are often designed for one type of noise distribution, e.g., additive white Gaussian noise. This means their performance is severely compromised in the presence of HF noise. Therefore, an AMC capable of mitigating the time-varying nature of HF noise is required. This article presents a robust AMC method for the classification of FSK, PSK, OQPSK, QAM, and amplitude-phase shift keying modulations in presence of HF noise using feature-based methods. Here, extracted features are insensitive to symbol synchronization and carrier frequency and phase offsets. The proposed AMC method is simple to implement as it uses decision-tree approach with pre-computed thresholds for signal classification. In addition, it is capable to classify type and order of modulation in both Gaussian and non-Gaussian environments.

Digital Modulation Classification via Sequential Monte Carlo for Frequency-Nonselective Fading MIMO Channels

Digital Modulation Classification via Sequential Monte Carlo for Frequency-Nonselective Fading MIMO Channels

A waveform digitizer-based automatic modulation classifier for a flexible spectrum management

The paper presents the development and validation of an automatic digital modulation classifier based on a waveform digitizer architecture. The proposed automatic digital modulation classifier can be used for signal recognition in the next generation telecommunication systems, characterized by a flexible and dynamic management of the radio spectrum. It has been implemented on a PC architecture, connected with two different types of waveform digitizers. The experimental validation has been carried out by means of actual telecommunication signals, even in presence of narrowband and wideband interference.