electronics

Article
Automatic Modulation Recognition Based on Deep-Learning
Features Fusion of Signal and Constellation Diagram
Hui Han 1,† , Zhijian Yi 2 , Zhigang Zhu 2, *, Lin Li 2 , Shuaige Gong 1,† , Bin Li 3 and Mingjie Wang 4

1 State Key Laboratory of Complex Electromagnetic Environment Effects on Electronics and Information
System (CEMEE), Luoyang 471003, China
2 School of Electronic Engineering, Xidian University, Xi’an 710071, China
3 Xi’an Satellite Control Center, Xi’an 710043, China
4 Academy For Network and Communications of CETC, Shijiazhuang 050081, China
* Correspondence: zgzhu@xidian.edu.cn; Tel.: +86-029-88202256
† These authors contributed equally to this work.

Abstract: In signal communication based on a non-cooperative communication system, the receiver is

an unlicensed third-party communication terminal, and the modulation parameters of the transmitter
signal cannot be predicted in advance. After the RF signal passes through the RF band-pass filter,
low noise amplifier, and image rejection filter, the intermediate frequency signal is obtained by
down-conversion, and then the IQ signal is obtained in the baseband by using the intermediate
frequency band-pass filter and down-conversion. In this process, noise and signal frequency offset are
inevitably introduced. As the basis of subsequent analysis and interpretation, modulation recognition
has important research value in this environment. The introduction of deep learning also brings new
feature mining tools. Based on this, this paper proposes a signal modulation recognition method
based on multi-feature fusion and constructs a deep learning network with a double-branch structure
to extract the features of IQ signal and multi-channel constellation, respectively. It is found that
through the complementary characteristics of different forms of signals, a more complete signal
feature representation can be constructed. At the same time, it can better alleviate the influence of
noise and frequency offset on recognition performance, and effectively improve the classification
accuracy of modulation recognition.
Citation: Han, H.; Yi, Z.; Zhu, Z.; Li,
L.; Gong, S.; Li, B.; Wang, M.
Automatic Modulation Recognition Keywords: constellation diagram; IQ feature; modulation recognition; multi-feature fusion
Based on Deep-Learning Features
Fusion of Signal and Constellation
Diagram. Electronics 2023, 12, 552.
https://doi.org/10.3390/ 1. Introduction
electronics12030552 Automatic modulation recognition (AMR) is a critical step in signal detection and
Academic Editor: Giovanni Crupi subsequent demodulation tasks [1], which aim to utilize prior information of signals to iden-
tify the modulation type in cognitive radio, electronic countermeasures, electromagnetic
Received: 12 December 2022 spectrum monitoring and other fields, etc. With the rapid development of communication
Revised: 12 January 2023
technology and the increasing complexity of electromagnetic environment, the modulation
Accepted: 14 January 2023
types of signals have become more complex and diverse, and often have the character-
Published: 20 January 2023
istics of low intercept probability. These situations put forward higher requirements for
modulation recognition.
Traditional AMR methods can be mainly divided into two categories: methods based
Copyright: © 2023 by the authors.
on likelihood function [2–4] and methods based on feature extraction [5–7]. The former
Licensee MDPI, Basel, Switzerland. makes use of the Bayesian minimum error judgment criterion [8] to ensure the excellence
This article is an open access article of the identification results. However, such methods are sensitive to model parameters
distributed under the terms and and lack generality and robustness [9]. The method based on feature extraction is based
conditions of the Creative Commons on the hand-crafted features, then on designing the backend classifier for final prediction.
Attribution (CC BY) license (https:// Features generally include instantaneous features, statistical features, spectral correlation
creativecommons.org/licenses/by/ features and transform domain features, etc. The traditional classifier adopts Decision
4.0/). Trees (DT) [10], K-Neareat Neighbor (KNN) [11], Support Vector Machine (SVM) [12],

Electronics 2023, 12, 552. https://doi.org/10.3390/electronics12030552 https://www.mdpi.com/journal/electronics

Electronics 2023, 12, 552 2 of 13

and Artificial Neural Network (ANN) [13]. Muller et al. used the SVM classifier for
recognition based on instantaneous and phase characteristics, effectively alleviating the
over-dependence of ANN classifier on training samples [14,15]. The methods based on
an instantaneous feature have the advantages of small-amount computation and simple
feature extraction process, which however is easy to be affected by channel noise and is
difficult to tackle with low SNR circumstance. Therefore, the high-order cumulant (HOC)
feature that is more robust to carrier offset and Gaussian noise is proposed. Mirarab et
al. [16] used the eighth-order cumulant feature to distinguish 8PSK and 16PSK signals,
the algorithm showed certain robustness to frequency offset. Wu et al. [17] provided a
new direction for solving the problem of signal modulation recognition in blind channels
by using the characteristics of fourth-order cumulants. In 2009, Orlic et al. [18] proposed
the AMC method based on sixth-order cumulant, which further improved the signal
recognition rate under real-world channel conditions. Wong et al. [19] proposed a new
scheme combining HOC with the Naive Bayes (NB) classifier. The FB method completes the
mapping of the signal sequence to different feature spaces by manually designing features,
which face a large number of arithmetic and signal form changes. At the same time, the
excessive dependence on features also makes the recognition effect of the FB method not
ideal in the current complex electromagnetic environment.
With the development of deep learning (DL) theory, the research of modulation recog-
nition methods based on DL gradually deepened and gradually replaced the traditional
signal modulation recognition algorithm. Such algorithms are usually data-driven [20]
and do not rely on design and extraction processes of complex hand-crafted features, but
build compact signal representations through convolutional neural networks (CNN), which
autonomically learn the feature information with high differentiation and robustness in the
samples by using a large number of labeled data sets as the input samples of the network
for training. This greatly improves the effectiveness of the extracted features and improves
the recognition performance. Therefore, a large number of excellent modulation recognition
methods have been derived using deep learning methods.
Compared with the traditional AMR method, the DL-based AMR methods include
three types of feature representation:
(1) Feature representation. Using the original signal information, the signal is pro-
cessed into a combination of a series of eigenvalues. Such as HOC characteristics, High-
Order Moment (HOM) characteristics and other statistical characteristics. Hassank et
al. [21] used HOM and Feedforward Neural Networks (FNN) to complete the AMR task.
Xie et al. [22] utilized the neural network to extract sixth-order cumulant of signals, and
achieved better recognition results under multipath and frequency offset conditions.
(2) Sequence representation. The signal is transformed into a vector, which is sent to
the corresponding neural network as a sequence feature. Features in this pipeline include
amplitude and phase sequence [23,24], IQ sequence [25,26] and FFT sequence [27], and so
on. Wang et al. [28] used the knowledge-sharing CNN model to complete the learning of the
common features of IQ samples under different Signal to Noise Ratio (SNR), improving the
modulation recognition performance and generalization under different noise conditions.
Liu et al. [29] combined CNN, gated recurrent unit and DNN to complete high-precision
modulation recognition tasks.
(3) Image representation. The signal is transformed into a two-dimensional image,
which includes constellation diagrams [30,31], time–frequency maps [32,33], cyclic spec-
trums [34,35], etc. Actually, the task is transformed into an image problem, and then the
classic image recognition algorithm is used for recognition. Yang et al. [36] used con-
stellation diagrams to train the network weights and realized the recognition of multiple
modulated signals under different noises. Ma et al. [37] designed a new AMR algorithm us-
ing a low-order cyclic spectrum and deep residual network, which can effectively suppress
impulse noise and extract discriminant features.
In addition to the feature-level AMR method, in view of network architectures, many
excellent network structures have been proposed. In [38] achieved better recognition perfor-
Electronics 2023, 12, 552 3 of 13

mance based on Constellation Graph Projection (GCP) algorithm and classical Deep Belief
Network (DBN). In [39] used three-channel gray constellation diagrams and GoogleNet to
effectively avoid the artificial feature design process and improve the classification accuracy
of the signal. In [40] explored the classification performance of various convolutional
neural networks (VGG16, VGG19, ResNet50, etc.). The VGG16 and VGG19 networks
are composed of 16 or 19 small-size convolutional layers and three fully connected lay-
ers, respectively. ResNet50 is implemented by a five-stage residual connection block. By
comparing the results, this paper finds the performance advantages of ResNet50 in signal
modulation recognition based on constellation diagrams.
The above methods use the sequence or image representation of the signal, combined
with the deep neural network to extract its features, and obtained better recognition results.
However, only by combining signal preprocessing and deep learning methods to improve
the existing algorithms, the following problems still remain to be solved: (1) Compared with
the IQ signal, the constellation diagram loses some information of the signal. Meanwhile,
the constellation diagram is sensitive to the frequency offset of the signal. When the signal
sequence length is short, the constellation diagram features extracted by the network have
poor robustness, and the classification results are easily affected. (2) Compared with the
constellation diagram, the IQ signal is susceptible to noise. When the SNR is low, the
signal feature extracted by the network lacks discrimination, and the classification result is
not stable.
To solve the above problems, this paper proposes a deep-learning AMR structure of
fusing IQ signal and constellation diagram, using dual branches to extract IQ signal and
constellation depth features, respectively. In the signal branch, deep-learning architecture is
utilized to map IQ signals to high-dimensional feature space. For the constellation diagram
branch, a three-channel constellation diagram mapping module is embedded, and the
feature is performed on each channel. Through the feature fusion of two branches, the
feature information of the IQ signal and constellation diagram is complementary. At the
same time, the algorithm can better alleviate the performance loss caused by noise and
frequency bias. When SNR is −10 dB and the maximum random frequency bias is 50 kHz,
the classification accuracy of 89% can be guaranteed.
The paper are organized as follows: Section 2 introduces the composition and internal
details of the proposed feature fusion network, including convolution kernel parameters,
step size, number, etc. In Section 3, parameters design and experiments of the proposed
method are performed. Section 4 gives the conclusions.

2. Methods
In order to realize the feature matching and fusion between the IQ signal and constel-
lation diagram, the IQ signal is used as the original input, and the projection process of the
constellation diagram is embedded into the network structure.
Figure 1 is the signal and constellation diagram fusion network (SCFNet). The pro-
posed model can be mainly divided into three parts: IQ feature branch, three-channel
constellation diagram feature branch, and feature fusion module. SCFNet takes the IQ
signal as the input and sends it to the dual-branch network structure for further processing.
The IQ branch takes the original IQ signal as input, six convolution-pooling layer
groups are followed for feature mapping, and high-dimensional features are fused for
the AMR. The constellation branch maps IQ signal to three-channel constellations by
the enhanced constellation diagram (ECD) [41] algorithm, as shown in C1 , C2 and C3 in
Figure 1. C1 is a linear mapping, which aims to map the constellation matrix C into an 8-bit
grayscale image. C2 is a logarithmic map, used to enhance the small values in C, designed
to suppress pulse interference. C3 is an exponential mapping, which is used to enhance
the large value in C to suppress background noise. The expressions of the three mapping
functions are as follows:
C − Cmin
 
C1 = f 1 (C ) = × 255 (1)
Cmax − Cmin
Electronics 2023, 12, 552 4 of 13

where Cmax and Cmin are the maximum and minimum values of matrix C, respectively.

C2 = f 2 (C ) = bα1 log2 ( β 1 f 1 (C ) + 1)c (2)

f 1 (C )
 
C3 = f 3 (C ) = α2 e β2
−1 (3)

where α1 , α2 , β 1 , and β 2 are predefined constants.

I/Q signal

Input
C1 Feature
Fusion

Transpose

Output

C2
1D 2D
Convolution Convolution

Average Max
Pooling Pooling

Fully
Connected C3

Figure 1. Signal modulation recognition method based on multi-feature fusion.

After obtaining three constellation diagrams, the correspondence between the I and
Q components is captured by one-dimensional convolution unidirectional sliding. For
constellation diagram C1 , one-dimensional convolution will slide from top to bottom to
obtain the change of I with Q. For constellation diagram C2 , it will be transposed to
exchange I and Q, so as to obtain the change of Q with I. Since the constellation diagram is
a two-dimensional representation of the signal, the two-dimensional convolution is also
used to extract the features of C3 . The above processing for the individual sub-branch of the
constellation diagram makes full use of the information of the three-channel constellation
diagram and improves the richness of features. The proposed method designs different
branch structures for different characteristics. The network structure and parameters of
each branch are introduced below.
For the C1 and C2 constellation diagrams, the branch network structure is shown in
Table 1. The input I/Q signal dimension is 1024 × 2, and the pixel size of the constellation
diagram is 64 × 64. It can be seen from Table 1 that the one-dimensional convolution
of the 3 × 64 convolution kernel can reduce the two-dimensional constellation diagram
features with an input size of 64 × 64 to one-dimensional features with a size of 64 × 1.
Subsequently, the small-size convolution kernel of 3 × 1 will be used for processing in
order to reduce the number of parameters. After the convolution of each layer, ReLU is
used as the activation function. Finally, the outputs of both C1 and C2 constellation diagram
branches are 16 × 1.
Electronics 2023, 12, 552 5 of 13

Table 1. The constellation diagram branch network structure of C1 and C2 .

Types Size/Step Output Size

1D Convolution 3 × 64/1 64 × 1
Average Pooling 2 × 1/2 32 × 1
1D Convolution 3 × 1/1 32 × 1
Average Pooling 2 × 1/2 16 × 1

Table 2 shows the network structure of the C3 constellation diagram branch. The branch
uses two-dimensional convolution and maximum pooling to design the corresponding net-
work structure, and uses a 3 × 3 small-size convolution kernel for feature extraction, thereby
reducing the number of parameters and eventually outputting 4 × 4 two-dimensional features.
The fusion module reshapes it into 16 × 1 high-dimensional features.

Table 2. The constellation diagram branch network structure of C3 .

Types Size/Step Output Size

2D Convolution 3 × 3/2 32 × 32
Max Pooling 2 × 2/2 16 × 16
2D Convolution 3 × 3/2 8×8
Max Pooling 2 × 2/2 4×4

Table 3 is the IQ signal branch network structure, which has six convolution-pooling
layers for feature extraction. In addition to the first layer, the real and imaginary parts of
the IQ signal is subjected to a bidirectional one-dimensional convolution. The size of the
remaining convolution kernels is set to 3 × 1, which can effectively reduce parameters
amounts and save the operation and time overhead of training. It can be seen from the
table that the original IQ signal is finally output as a 16 × 1 high-dimensional feature
after dimensionality reduction, which is fused with the high-dimensional constellation
diagram features.

Table 3. The constellation diagram branch network structure of IQ signal.

Types Size/Step Output Size

1D Convolution 3 × 3/1 1024 × 1
Average Pooling 2 × 1/2 512 × 1
1D Convolution 3 × 1/1 512 × 1
Average Pooling 2 × 1/2 256 × 4
1D Convolution 3 × 1/1 256 × 1
Average Pooling 2 × 1/2 128 × 1
1D Convolution 3 × 1/1 128 × 1
Average Pooling 2 × 1/2 64 × 1
1D Convolution 3 × 1/1 64 × 1
Average Pooling 2 × 1/2 32 × 1
1D Convolution 3 × 1/1 32 × 1
Average Pooling 2 × 1/2 16 × 1

As shown in Figure 1, the fused features are sent to the convolution-pooling assembly
and two fully connected layers for final classification. Finally, the modulation type of the
output signal is obtained. Because multi-layer convolution has been included in the branch
before fusion, the fused features are not trained by the too deep convolutional neural
network.

3. Experiments
In order to study the effectiveness of the proposed method based on multi-feature
fusion, this section will design the optimal parameters through ablation experiments.
Electronics 2023, 12, 552 6 of 13

Meanwhile, the performance of the feature fusion network is explored by the single feature
following the classical feature extraction network. In the experiment, the classification
accuracy index is used to measure the recognition effect, which is defined as follows:

Nture
accuracy = × 100% (4)
Nsignal

where Nture and Nsignal represent the number of signals correctly identified in the test set
and the total number of signals, respectively.

3.1. Detailed Overview of Network Training

This section will design the parameters of the network from the aspects of dataset size
and training learning rate. The proposed method is used to identify 4ASK, 2PSK, 4PSK,
8PSK, 16QAM, 32QAM, 64QAM and 128QAM signals, and the optimal parameters are
adjusted and determined. The initial network learning rate is set to 0.001, the batch size
is 64, the optimizer is Adam, and the IQ signal length is 1024 × 2. The total number of
training epochs is 200, and the training time is 132.7 s.

3.1.1. Recognition Comparisons of the Proposed Methods under Various Signal Amounts
To study the influence of different dataset sizes on recognition accuracy, three datasets
with different sizes are produced in this section. The dataset contains eight types of
signals under five SNRs. Due to the complex channel environment in the actual signal
transmission process, the received signal may have a frequency offset. In order to simulate
the real transmission scenario, the maximum random frequency offset of the signal is set
to 50 kHz. At each SNR, the number of samples for each type of signal is a fixed value
of 1000. The number of single signal samples in the training set is set to 250, 500 and 750,
respectively, and the rest is used as the test data set, that is, the three data sets contain
10,000, 20,000, and 30,000 signal samples, respectively. The simulation results are shown in
Figure 2.

Figure 2. The influence of different signal amounts on classification accuracy.

It can be seen from Figure 2 that the model based on dataset 3 training has the best
classification performance and can obtain higher recognition accuracy. Due to the existence
of noise, digital signals will have different degrees of error codes, which brings difficulties to
Electronics 2023, 12, 552 7 of 13

subsequent modulation recognition. The effectiveness of features can alleviate the influence
of noise to a certain extent. The more samples in the dataset, the greater the possibility of
extracting effective features, and the better the trained model. The results in the figure also
reflect this conclusion.

3.1.2. The Influence of Learning Rate on the Accuracy Performance

The learning rate is an important hyper-parameter in the training process of the deep
neural network, which also affects the training effect of the model. In order to select the
optimal learning rate, this section sets different learning rates. The simulation results are
shown in Figure 3. As seen in Figure 3, different learning rates will lead to a big gap in
classification accuracy. When the learning rate is 0.001, the model has the best classification
performance. Too large or too small a learning rate will cause different degrees of accuracy
loss. When the learning rate is too large, the network may not converge; when it is too
small, the network may converge slowly or fall into the local optimum, and there is not a
simple linear relationship between learning rate and classification accuracy.

Figure 3. The influence of different learning rates on classification accuracy.

3.2. Analysis of Experimental Results

3.2.1. Ablation Experiment
The modulation recognition algorithm proposed in this paper aims to improve the clas-
sification effect and anti-disturbance performance of the network by using complementary
information between IQ signal features and constellation diagram features. To study the
effectiveness of the proposed method and the fusion features, the following experiments
will be carried out for different scenarios. The input dataset contains eight modulation
signals under 11 SNRs, the training set and the test set contain 30,000 and 10,000 signals,
respectively, and the IQ signal length is set to 1024 × 2.
Figure 4 is the accuracy curve of the modulation recognition method based on different
features on the test set without frequency offset. The result shows that the classification
accuracy of the recognition method based on joint features and constellation diagram is
significantly superior to that of the method based on IQ signal. Under the set SNRs, the
classification accuracy based on joint features is higher than that based on the constellation
diagram. When the SNR is −10 dB, the accuracy of the method based on the IQ signal
is only 75.5%, while the classification accuracy of the other two methods is more than
Electronics 2023, 12, 552 8 of 13

98%. Under the condition of no frequency offset, compared with the original IQ signal, the
constellation diagram feature has a better characterization of the signal, which is conducive
to the modulation classification of the signal.

Figure 4. The influence of different features on classification accuracy without frequency offset.

To explore the influence of signal frequency offset on the performance of the proposed
method and the traditional single-feature modulation recognition algorithm, an experi-
mental scene with a maximum random frequency offset of 50 kHz is designed. There are
eight modulation signals under 5 SNRs in the target dataset, the training set and the test
set contain 30,000 and 10,000 signals, respectively, and the final experimental results are
shown in Figure 5. It shows that the classification accuracy based on different features is
reduced to some extent in the scene with frequency offset. When the SNR is −10 dB, the
classification accuracy based on the single constellation diagram feature is reduced from
98% to 81%. In the presence of frequency offset, the constellation diagram point clusters
will rotate to different degrees by making the received signal lose the original constellation
position information on the IQ plane, which can identify not only the signal modulation
mode well, but suffer poor recognition ability of the algorithm. Constellation diagram
features are easily affected by frequency offset. Under the condition of high frequency
offset, the recognition ability of the algorithm is poor. However, under various SNRs, the
accuracy of the modulation recognition method based on joint features is higher than that of
the other two methods based on single features. Thus, the joint feature effectively combines
the advantages of two single features and can obtain better recognition results.
Electronics 2023, 12, 552 9 of 13

Figure 5. The influence of different features on classification accuracy when the maximum random
frequency offset is 50 kHz.

3.2.2. Contrast Experiments

After selecting the optimal parameters and verifying the effectiveness of fusion fea-
tures, to further analyze the overall performance of the proposed method, this section
compares the proposed method with other advanced network models applied to mod-
ulation recognition through simulation experiments according to the selected optimal
parameters. The comparison methods include the classical convolutional neural network
models GoogleNet, VGGNet, GCP–DBN, and so on.
The experiment uses 8 kinds of modulation signals under 5 kinds of SNRs as classi-
fication targets. Each type of signal in the data set has 1000 signal samples. The ratio of
train set to test set is 3:1, and other parameters are set as shown in the Table 4, Fo and Fc
represent the maximum random frequency offset and carrier frequency parameters of the
signal, respectively.

Table 4. Experimental parameter settings.

Parameters Value
SNR Range −10 dB~10 dB
Sampling Frequency 5 GHz
Signal Length 1024 × 2
Learning Rate 0.001
Fo/Fc 0.074

Considering the classification accuracy of various methods on the test set for 8 com-
mon modulation signals of 4ASK, 2PSK, 4PSK, 8PSK, 16QAM, 32QAM, 64QAM, and
128QAM under the condition of frequency offset, the results are shown in the Figure 6,
some experimental results are shown in Table 5.
Electronics 2023, 12, 552 10 of 13

Table 5. Classification accuracy with maximum random frequency offset of 50 kHz.

Networks 0dB −5 dB −10 dB

SCFNet 98.2% 94.6% 89.5%
GoogleNet 95.7% 90.0% 82.0%
VGGNet 93.6% 92.1% 89.6%
GCP-DBN 89.0% 73.0% 60.9%

Figure 6. Classification accuracy of different methods when the maximum random frequency offset
is 50 kHz.

The figure shows that the recognition performance of the GCP–DBN algorithm is more
unstable. As the SNR gradually decreases, its classification accuracy decreases rapidly. At
−10 dB, the classification accuracy is only about 60%, while the classification accuracy of
the proposed method is 89%, which is much higher than GCP–DBN algorithm. Comparing
the classification accuracy of different methods, it can be found that the classification rate
of the proposed SCFNet method is higher than other methods. At the same time, compared
with other network structures, the structure of SCFNet is simpler with fewer parameters
and calculations. When the signal sample size in the dataset is large, the corresponding
training time overhead can be reduced. Under the same conditions, SCFNet has better
recognition performance.
To study the influence of frequency offset on the experimental results, and to facilitate
the comparison of the anti-frequency offset performance of different methods, five maxi-
mum random frequency offsets of 0 kHz, 25 kHz, 50 kHz and 100 kHz are set when the
SNR is −5 dB, and different datasets are made. The dataset composition is consistent with
the previous experiment. The experimental results of the test set are shown in Figure 7.
The HOC1 in the experiment is four high-order cumulant features of C42, |C40|/|C42|,
|C63|2 /|C42|3 and |C80|/|C42|2 . HOC2 represents the five high order cumulant features
of C40, C42, C43, C60 and C63. Under the above commonly used HOC features, the
modulation signals selected in the experiment can be better distinguished The HOC1–SVM
method uses the support vector machine as the classifier and maps the input data to the
high-dimensional space through the kernel function, thereby establishing the maximum
interval hyperplane, and classifying by maximizing the distance between the sample and
the decision surface. HOC2–NaiveBayes utilizes the NaiveBayes classifier to output the
final classification result using the optimal criterion.
Electronics 2023, 12, 552 11 of 13

Figure 7. Classification accuracy of different algorithms under multiple frequency offsets.

The figure shows that in the scene without frequency offset, the classification accuracy
of GoogleNet, VGGNet and SCFNet is similar and the recognition effect is better, while
the classification accuracy of GCP–DBN, HOC1–SVM and HOC2–NaiveBayes is lower.
With the increase of frequency offset, the classification accuracy of all algorithms has been
reduced to varying degrees. In the same frequency offset scenario, the accuracy of GCP–
DBN algorithm is generally lower than other algorithms, and the classification accuracy
of the modulation recognition method based on SVM is slightly lower than that based on
NaiveBayes. Combining Figures 6 and 7, it can be found that the performance of GCP–DBN
method is relatively unstable, and its classification accuracy will be greatly affected in
the face of common interference in the process of signal transmissions, such as noise and
frequency offset. For the three methods of GoogleNet, VGGNet and SCFNet, when the
maximum random frequency offset is in the range of 25 kHz to 100 kHz, GoogleNet and
VGGNet both decrease by more than 15%, while the classification accuracy of SCFNet
decreases from 96% to 90%. It can be seen that the large frequency offset has a great
influence on the recognition performance of various networks. However, SCFNet can
alleviate the performance loss caused by noise and frequency offset, and has excellent
modulation recognition performance in non-ideal large interference scenarios.

4. Conclusions
The experimental results show that the proposed method is with superior recognition
performance under the conditions of low SNR and high frequency offset. In our scenario,
SCFNet achieves the recognition rate of 90% to 95%. When the maximum frequency offset
increases from 25 kHz to 100 kHz, the recognition accuracy of the classical convolutional
neural network is reduced by at least 15%, while the classification accuracy of the proposed
method only has a decrease within 6%. It can be seen that SCFNet has stronger robustness
to the influence of noise and frequency offset. The main reason is that the multi-feature
fusion network combines the feature information of IQ signal and constellation diagram,
which improves the robustness of extracted features. Additionally, combination of signal
information and constellation diagram representation enables to alleviate the influence of
frequency offset and noise on recognition, which provides a new research idea to AMR.
Considering that different types of single features are complementary, future work can
focus on the fusion of more features, in order to alleviate the information loss during signal
transmission and demodulation in the current complex electromagnetic environment.
Electronics 2023, 12, 552 12 of 13

Author Contributions: Conceptualization, Z.Z. and L.L.; methodology, Z.Z. and Z.Y.; formal analysis,
L.L. and M.W.; funding acquisition, S.G., H.H., Z.Z. and L.L.; investigation, B.L.; supervision, L.L.;
validation, Z.Y. and B.L.; visualization, S.G., Z.Z. and B.L.; writing—original draft, Z.Y. and Z.Z.;
writing—review and editing, Z.Z. and L.L. All authors have read and agreed to the published version
of the manuscript.
Funding: This work is partially supported by the Open Project of State Key Laboratory of Com-
plex Electromagnetic Environment Effects on Electronics and Information System under Grants ]
CEMEE2022K0103A, the National Natural Science Foundation of China under Grants ] 62203343, ]
62071349 and ] U21A20455, and Key Research and Development Program of Shaanxi (Program No.
2023-YBGY-223).
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement: Not applicable.
Acknowledgments: The authors wish to express their appreciation to the editors for their rigorous
and efficient work and the reviewers for their helpful suggestions, which greatly improved the
presentation of this paper.
Conflicts of Interest: The authors declare that they have no conflict of interest to report regarding
the present study.

References
1. Mao, Q.; Hu, F.; Hao, Q. Deep learning for intelligent wireless networks: A comprehensive survey. IEEE Commun. Surv. Tutor.
2018, 20, 2595–2621. [CrossRef]
2. Häring, L.; Chen, Y.; Czylwik, A. Automatic modulation classification methods for wireless OFDM systems in TDD mode. IEEE
Trans. Commun. 2010, 58, 2480–2485. [CrossRef]
3. Salam, A.O.A.; Sheriff, R.E.; Al-Araji, S.R.; Mezher, K.; Nasir, Q. A unified practical approach to modulation classification in
cognitive radio using likelihood-based techniques. In Proceedings of the 2015 IEEE 28th Canadian Conference on Electrical and
Computer Engineering (CCECE), Halifax, NS, Canada, 3–6 May 2015; pp. 1024–1029.
4. Xu, J.L.; Su, W.; Zhou, M. Likelihood-ratio approaches to automatic modulation classification. IEEE Trans. Syst. Man Cybern. Part
C (Appl. Rev.) 2010, 41, 455–469. [CrossRef]
5. Huang, S.; Lin, C.; Xu, W.; Gao, Y.; Feng, Z.; Zhu, F. Identification of active attacks in internet of things: Joint model-and
data-driven automatic modulation classification approach. IEEE Internet Things J. 2020, 8, 2051–2065. [CrossRef]
6. Dobre, O.A.; Abdi, A.; Bar-Ness, Y.; Su, W. Survey of automatic modulation classification techniques: Classical approaches and
new trends. IET Commun. 2007, 1, 137–156. [CrossRef]
7. Huang, S.; Yao, Y.; Wei, Z.; Feng, Z.; Zhang, P. Automatic modulation classification of overlapped sources using multiple
cumulants. IEEE Trans. Veh. Technol. 2016, 66, 6089–6101. [CrossRef]
8. Abdi, A.; Dobre, O.A.; Choudhry, R.; Bar-Ness, Y.; Su, W. Modulation classification in fading channels using antenna arrays. In
Proceedings of the IEEE MILCOM 2004. Military Communications Conference, Monterey, CA, USA, 31 October–3 November
2004; Volume 1, pp. 211–217.
9. Wei, W.; Mendel, J.M. Maximum-likelihood classification for digital amplitude-phase modulations. IEEE Trans. Commun. 2000,
48, 189–193. [CrossRef]
10. Furtado, R.S.; Torres, Y.P.; Silva, M.O.; Colares, G.S.; Pereira, A.M.; Amoedo, D.A.; Valad ao, M.D.; Carvalho, C.B.; da Costa, A.L.;
Júnior, W.S. Automatic Modulation Classification in Real Tx/Rx Environment using Machine Learning and SDR. In Proceedings
of the 2021 IEEE International Conference on Consumer Electronics (ICCE), Las Vegas, NV, USA, 10–12 January 2021; pp. 1–4.
11. Zhu, Z.; Aslam, M.W.; Nandi, A.K. Augmented genetic programming for automatic digital modulation classification. In
Proceedings of the 2010 IEEE International Workshop on Machine Learning for Signal Processing, Kittila, Finland, 29 August–1
September 2010; pp. 391–396.
12. Dong, S.; Li, Z.; Zhao, L. A modulation recognition algorithm based on cyclic spectrum and SVM classification. In Proceedings of
the 2020 IEEE 4th Information Technology, Networking, Electronic and Automation Control Conference (ITNEC), Chongqing,
China, 12–14 June 2020; Volume 1, pp. 2123–2127.
13. Ya, T.; Lin, Y.; Wang, H. Modulation recognition of digital signal based on deep auto-ancoder network. In Proceedings of the 2017
IEEE International Conference on Software Quality, Reliability and Security Companion (QRS-C), Prague, Czech Republic, 25–29
July 2017; pp. 256–260.
14. Muller, F.C.; Cardoso, C.; Klautau, A. A front end for discriminative learning in automatic modulation classification. IEEE
Commun. Lett. 2011, 15, 443–445. [CrossRef]
15. Nandi, A.K.; Azzouz, E.E. Algorithms for automatic modulation recognition of communication signals. IEEE Trans. Commun.
1998, 46, 431–436. [CrossRef]
Electronics 2023, 12, 552 13 of 13

16. Mirarab, M.; Sobhani, M. Robust modulation classification for PSK/QAM/ASK using higher-order cumulants. In Proceedings of the
2007 6th International Conference on Information, Communications & Signal Processing, Singapore, 10–13 December 2007; pp. 1–4.
17. Wu, H.C.; Saquib, M.; Yun, Z. Novel automatic modulation classification using cumulant features for communications via
multipath channels. IEEE Trans. Wirel. Commun. 2008, 7, 3098–3105.
18. Orlic, V.D.; Dukic, M.L. Automatic modulation classification algorithm using higher-order cumulants under real-world channel
conditions. IEEE Commun. Lett. 2009, 13, 917–919. [CrossRef]
19. Wong, M.D.; Ting, S.K.; Nandi, A.K. Naive Bayes classification of adaptive broadband wireless modulation schemes with higher
order cumulants. In Proceedings of the 2008 2nd International Conference on Signal Processing and Communication Systems,
Gold Coast, QLD, Australia, 15–17 December 2008; pp. 1–5.
20. Zhu, Z.; Yi, Z.; Li, S.; Li, L. Deep muti-modal generic representation auxiliary learning networks for end-to-end radar emitter
classification. Aerospace 2022, 9, 732. [CrossRef]
21. Hassan, K.; Dayoub, I.; Hamouda, W.; Berbineau, M. Automatic modulation recognition using wavelet transform and neural
network. In Proceedings of the 2009 9th International Conference on Intelligent Transport Systems Telecommunications (ITST),
Lille, France, 20–22 October 2009; pp. 234–238.
22. Xie, W.; Hu, S.; Yu, C.; Zhu, P.; Peng, X.; Ouyang, J. Deep learning in digital modulation recognition using high order cumulants.
IEEE Access 2019, 7, 63760–63766. [CrossRef]
23. Hong, S.; Zhang, Y.; Wang, Y.; Gu, H.; Gui, G.; Sari, H. Deep learning-based signal modulation identification in OFDM systems.
IEEE Access 2019, 7, 114631–114638. [CrossRef]
24. Zhang, Z.; Luo, H.; Wang, C.; Gan, C.; Xiang, Y. Automatic modulation classification using CNN-LSTM based dual-stream
structure. IEEE Trans. Veh. Technol. 2020, 69, 13521–13531. [CrossRef]
25. O’Shea, T.J.; Corgan, J.; Clancy, T.C. Convolutional radio modulation recognition networks. In Proceedings of the International
Conference on Engineering Applications of Neural Networks; Springer: Cham, Switzerland, 2016; pp. 213–226.
26. Zheng, S.; Qi, P.; Chen, S.; Yang, X. Fusion methods for CNN-based automatic modulation classification. IEEE Access 2019, 7,
66496–66504. [CrossRef]
27. Kulin, M.; Kazaz, T.; Moerman, I.; De Poorter, E. End-to-end learning from spectrum data: A deep learning approach for wireless
signal identification in spectrum monitoring applications. IEEE Access 2018, 6, 18484–18501. [CrossRef]
28. Wang, Y.; Gui, G.; Ohtsuki, T.; Adachi, F. Multi-task learning for generalized automatic modulation classification under
non-Gaussian noise with varying SNR conditions. IEEE Trans. Wirel. Commun. 2021, 20, 3587–3596. [CrossRef]
29. Li, R.; Li, L.; Yang, S.; Li, S. Robust automated VHF modulation recognition based on deep convolutional neural networks. IEEE
Commun. Lett. 2018, 22, 946–949. [CrossRef]
30. Wang, Y.; Liu, M.; Yang, J.; Gui, G. Data-driven deep learning for automatic modulation recognition in cognitive radios. IEEE
Trans. Veh. Technol. 2019, 68, 4074–4077. [CrossRef]
31. Li, L.; Dong, Z.; Zhu, Z.; Jiang, Q. Deep-learning hopping capture model for automatic modulation classification of wireless
communication signals. IEEE Trans. Aerosp. Electron. Syst. 2022.
32. Zhu, Z.; Ji, H.; Zhang, W.; Li, L.; Ji, T. Complex convolutional neural network for signal representation and its application to
radar emitter recognition. IEEE Commun. Lett. 2023. [CrossRef]
33. Zeng, Y.; Zhang, M.; Han, F.; Gong, Y.; Zhang, J. Spectrum analysis and convolutional neural network for automatic modulation
recognition. IEEE Wirel. Commun. Lett. 2019, 8, 929–932. [CrossRef]
34. Yan, X.; Liu, G.; Wu, H.C.; Feng, G. New automatic modulation classifier using cyclic-spectrum graphs with optimal training
features. IEEE Commun. Lett. 2018, 22, 1204–1207. [CrossRef]
35. Cao, S.; Zhang, W. Carrier frequency and symbol rate estimation based on cyclic spectrum. J. Syst. Eng. Electron. 2020, 31, 37–44.
[CrossRef]
36. Yang, Z.; Zhang, L. Modulation Classification Based on Signal Constellation and Yolov3. In Proceedings of the 2021 4th International
Conference on Information Communication and Signal Processing (ICICSP), Shanghai, China, 24–26 September 2021; pp. 143–146.
37. Ma, J.; Jiang, F. Automatic Modulation Classification Using Fractional Low Order Cyclic Spectrum and Deep Residual Networks
in Impulsive Noise. In Proceedings of the 2021 IEEE MTT-S International Wireless Symposium (IWS), Nanjing, China, 23–26 May
2021; pp. 1–3.
38. Mendis, G.J.; Wei-Kocsis, J.; Madanayake, A. Deep learning based radio-signal identification with hardware design. IEEE Trans.
Aerosp. Electron. Syst. 2019, 55, 2516–2531. [CrossRef]
39. Peng, S.; Jiang, H.; Wang, H.; Alwageed, H.; Zhou, Y.; Sebdani, M.M.; Yao, Y.D. Modulation classification based on signal
constellation diagrams and deep learning. IEEE Trans. Neural Netw. Learn. Syst. 2018, 30, 718–727. [CrossRef]
40. Tian, X.; Chen, C. Modulation pattern recognition based on Resnet50 neural network. In Proceedings of the 2019 IEEE 2nd International
Conference on Information Communication and Signal Processing (ICICSP), Weihai, China, 28–30 September 2019; pp. 34–38.
41. Li, L.; Li, M.; Zhu, Z.; Li, S.; Dai, C. An Efficient Digital Modulation Classification Method Using the Enhanced Constellation Dia-
gram. Available online: https://www.researchgate.net/publication/365374312_An_Efficient_Digital_Modulation_Classification_
Method_Using_the_Enhanced_Constellation_Diagram (accessed on 17 November 2022).

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.