In recent years,deep learning-based signal recognition technology has gained attention and emerged as an important approach for safeguarding the electromagnetic environment.However,training deep learning-based classif...In recent years,deep learning-based signal recognition technology has gained attention and emerged as an important approach for safeguarding the electromagnetic environment.However,training deep learning-based classifiers on large signal datasets with redundant samples requires significant memory and high costs.This paper proposes a support databased core-set selection method(SD)for signal recognition,aiming to screen a representative subset that approximates the large signal dataset.Specifically,this subset can be identified by employing the labeled information during the early stages of model training,as some training samples are labeled as supporting data frequently.This support data is crucial for model training and can be found using a border sample selector.Simulation results demonstrate that the SD method minimizes the impact on model recognition performance while reducing the dataset size,and outperforms five other state-of-the-art core-set selection methods when the fraction of training sample kept is less than or equal to 0.3 on the RML2016.04C dataset or 0.5 on the RML22 dataset.The SD method is particularly helpful for signal recognition tasks with limited memory and computing resources.展开更多
With the advent of the Industry 5.0 era,the Internet of Things(IoT)devices face unprecedented proliferation,requiring higher communications rates and lower transmission delays.Considering its high spectrum efficiency,...With the advent of the Industry 5.0 era,the Internet of Things(IoT)devices face unprecedented proliferation,requiring higher communications rates and lower transmission delays.Considering its high spectrum efficiency,the promising filter bank multicarrier(FBMC)technique using offset quadrature amplitude modulation(OQAM)has been applied to Beyond 5G(B5G)industry IoT networks.However,due to the broadcasting nature of wireless channels,the FBMC-OQAMindustry IoT network is inevitably vulnerable to adversary attacks frommalicious IoT nodes.The FBMC-OQAMindustry cognitive radio network(ICRNet)is proposed to ensure security at the physical layer to tackle the above challenge.As a pivotal step of ICRNet,blind modulation recognition(BMR)can detect and recognize the modulation type of malicious signals.The previous works need to accomplish the BMR task of FBMC-OQAM signals in ICRNet nodes.A novel FBMC BMR algorithm is proposed with the transform channel convolution network(TCCNet)rather than a complicated two-dimensional convolution.Firstly,this is achieved by designing a low-complexity binary constellation diagram(BCD)gridding matrix as the input of TCCNet.Then,a transform channel convolution strategy is developed to convert the image-like BCD matrix into a serieslike data format,accelerating the BMR process while keeping discriminative features.Monte Carlo experimental results demonstrate that the proposed TCCNet obtains a performance gain of 8%and 40%over the traditional inphase/quadrature(I/Q)-based and constellation diagram(CD)-based methods at a signal noise ratio(SNR)of 12 dB,respectively.Moreover,the proposed TCCNet can achieve around 29.682 and 2.356 times faster than existing CD-Alex Network(CD-AlexNet)and I/Q-Convolutional Long Deep Neural Network(I/Q-CLDNN)algorithms,respectively.展开更多
This paper proposes a novel open set recognition method,the Spatial Distribution Feature Extraction Network(SDFEN),to address the problem of electromagnetic signal recognition in an open environment.The spatial distri...This paper proposes a novel open set recognition method,the Spatial Distribution Feature Extraction Network(SDFEN),to address the problem of electromagnetic signal recognition in an open environment.The spatial distribution feature extraction layer in SDFEN replaces convolutional output neural networks with the spatial distribution features that focus more on inter-sample information by incorporating class center vectors.The designed hybrid loss function considers both intra-class distance and inter-class distance,thereby enhancing the similarity among samples of the same class and increasing the dissimilarity between samples of different classes during training.Consequently,this method allows unknown classes to occupy a larger space in the feature space.This reduces the possibility of overlap with known class samples and makes the boundaries between known and unknown samples more distinct.Additionally,the feature comparator threshold can be used to reject unknown samples.For signal open set recognition,seven methods,including the proposed method,are applied to two kinds of electromagnetic signal data:modulation signal and real-world emitter.The experimental results demonstrate that the proposed method outperforms the other six methods overall in a simulated open environment.Specifically,compared to the state-of-the-art Openmax method,the novel method achieves up to 8.87%and 5.25%higher micro-F-measures,respectively.展开更多
Interference signals recognition plays an important role in anti-jamming communication.With the development of deep learning,many supervised interference signals recognition algorithms based on deep learning have emer...Interference signals recognition plays an important role in anti-jamming communication.With the development of deep learning,many supervised interference signals recognition algorithms based on deep learning have emerged recently and show better performance than traditional recognition algorithms.However,there is no unsupervised interference signals recognition algorithm at present.In this paper,an unsupervised interference signals recognition method called double phases and double dimensions contrastive clustering(DDCC)is proposed.Specifically,in the first phase,four data augmentation strategies for interference signals are used in data-augmentation-based(DA-based)contrastive learning.In the second phase,the original dataset’s k-nearest neighbor set(KNNset)is designed in double dimensions contrastive learning.In addition,a dynamic entropy parameter strategy is proposed.The simulation experiments of 9 types of interference signals show that random cropping is the best one of the four data augmentation strategies;the feature dimensional contrastive learning in the second phase can improve the clustering purity;the dynamic entropy parameter strategy can improve the stability of DDCC effectively.The unsupervised interference signals recognition results of DDCC and five other deep clustering algorithms show that the clustering performance of DDCC is superior to other algorithms.In particular,the clustering purity of our method is above 92%,SCAN’s is 81%,and the other three methods’are below 71%when jammingnoise-ratio(JNR)is−5 dB.In addition,our method is close to the supervised learning algorithm.展开更多
In the past ten years, many high-quality datasets have been released to support the rapid development of deep learning in the fields of computer vision, voice, and natural language processing. Nowadays, deep learning ...In the past ten years, many high-quality datasets have been released to support the rapid development of deep learning in the fields of computer vision, voice, and natural language processing. Nowadays, deep learning has become a key research component of the Sixth-Generation wireless systems(6G) with numerous regulatory and defense applications. In order to facilitate the application of deep learning in radio signal recognition, in this work, a large-scale real-world radio signal dataset is created based on a special aeronautical monitoring system-Automatic Dependent Surveillance-Broadcast(ADS-B). This paper makes two main contributions. First, an automatic data collection and labeling system is designed to capture over-the-air ADS-B signals in the open and real-world scenario without human participation. Through data cleaning and sorting, a high-quality dataset of ADS-B signals is created for radio signal recognition. Second, we conduct an in-depth study on the performance of deep learning models using the new dataset, as well as comparison with a recognition benchmark using machine learning and deep learning methods.Finally, we conclude this paper with a discussion of open problems in this area.展开更多
The optical fiber pre-waming system (OFPS) has been gradually considered as one of the important means for pipeline safety monitoring. Intrusion signal types are correctly identified which could reduce the cost of t...The optical fiber pre-waming system (OFPS) has been gradually considered as one of the important means for pipeline safety monitoring. Intrusion signal types are correctly identified which could reduce the cost of troubleshooting and maintenance of the pipeline. Most of the previous feature extraction methods in OFPS are usually quested from the view of time domain. However, in some cases, there is no distinguishing feature in the time domain. In the paper, firstly, the intrusion signal features of the running, digging, and pick mattock are extracted in the frequency domain by multi-level wavelet decomposition, that is, the intrusion signals are decomposed into five bands. Secondly, the average energy ratio of different frequency bands is obtained, which is considered as the feature of each intrusion type. Finally, the feature samples are sent into the random vector functional-link (RVFL) network for training to complete the classification and identification of the signals. Experimental results show that the algorithm can correctly distinguish the different intrusion signals and achieve higher recognition rate.展开更多
In this paper,we investigate the problem of key radar signal sorting and recognition in electronic intelligence(ELINT).Our major contribution is the development of a combined approach based on clustering and pulse rep...In this paper,we investigate the problem of key radar signal sorting and recognition in electronic intelligence(ELINT).Our major contribution is the development of a combined approach based on clustering and pulse repetition interval(PRI)transform algorithm,to solve the problem that the traditional methods based on pulse description word(PDW)were not exclusively targeted at tiny particular signals and were less time-efficient.We achieve this in three steps:firstly,PDW presorting is carried out by the DBSCAN(Density-Based Spatial Clustering of Applications with Noise)clustering algorithm,and then PRI estimates of each cluster are obtained by the PRI transform algorithm.Finally,by judging the matching between various PRI estimates and key targets,it is determined whether the current signal contains key target signals or not.Simulation results show that the proposed method should improve the time efficiency of key signal recognition and deal with the complex signal environment with noise interference and overlapping signals.展开更多
We used data from the Xinjiang Digital Seismic Network and PSD( Power Spectral Density) method to perform noise level assessment for six stations. We calculated the median of the power spectral density to evaluate the...We used data from the Xinjiang Digital Seismic Network and PSD( Power Spectral Density) method to perform noise level assessment for six stations. We calculated the median of the power spectral density to evaluate the noise level of different stations. After the comparison of the power spectral density of different stations and the airgun signal recognition,we found that noise level of stations with recognizable airgun source signals is lower than that of stations without recognizable signals. The largest difference of the power spectral density is 40 d B,and the smallest is 15 d B. Finally,we found that the failure to recognize the signal of airgun sources at some stations is due to the noise level.展开更多
In this Letter, a method based on the effects of imperfect oscillators in lasers is proposed to distinguish targets in continuous wave tracking lidar. This technique is based on the fact that each lidar signal source ...In this Letter, a method based on the effects of imperfect oscillators in lasers is proposed to distinguish targets in continuous wave tracking lidar. This technique is based on the fact that each lidar signal source has a specific influence on the phase noise that makes real targets from the false ones. A simulated signal is produced by complex circuits, modulators, memory, and signal oscillators. For example, a deception laser beam has an unequal and variable phase noise from a real target. Thus, the phase noise of transmitted and received signals does not have the same power levels and patterns. To consider the performance of the suggested method, the probability of detection(PD) is shown for various signal-to-noise ratios and signal-to-jammer ratios based on experimental outcomes.展开更多
基金supported by National Natural Science Foundation of China(62371098)Natural Science Foundation of Sichuan Province(2023NSFSC1422)+1 种基金National Key Research and Development Program of China(2021YFB2900404)Central Universities of South west Minzu University(ZYN2022032).
文摘In recent years,deep learning-based signal recognition technology has gained attention and emerged as an important approach for safeguarding the electromagnetic environment.However,training deep learning-based classifiers on large signal datasets with redundant samples requires significant memory and high costs.This paper proposes a support databased core-set selection method(SD)for signal recognition,aiming to screen a representative subset that approximates the large signal dataset.Specifically,this subset can be identified by employing the labeled information during the early stages of model training,as some training samples are labeled as supporting data frequently.This support data is crucial for model training and can be found using a border sample selector.Simulation results demonstrate that the SD method minimizes the impact on model recognition performance while reducing the dataset size,and outperforms five other state-of-the-art core-set selection methods when the fraction of training sample kept is less than or equal to 0.3 on the RML2016.04C dataset or 0.5 on the RML22 dataset.The SD method is particularly helpful for signal recognition tasks with limited memory and computing resources.
基金supported by the National Natural Science Foundation of China(Nos.61671095,61371164)the Project of Key Laboratory of Signal and Information Processing of Chongqing(No.CSTC2009CA2003).
文摘With the advent of the Industry 5.0 era,the Internet of Things(IoT)devices face unprecedented proliferation,requiring higher communications rates and lower transmission delays.Considering its high spectrum efficiency,the promising filter bank multicarrier(FBMC)technique using offset quadrature amplitude modulation(OQAM)has been applied to Beyond 5G(B5G)industry IoT networks.However,due to the broadcasting nature of wireless channels,the FBMC-OQAMindustry IoT network is inevitably vulnerable to adversary attacks frommalicious IoT nodes.The FBMC-OQAMindustry cognitive radio network(ICRNet)is proposed to ensure security at the physical layer to tackle the above challenge.As a pivotal step of ICRNet,blind modulation recognition(BMR)can detect and recognize the modulation type of malicious signals.The previous works need to accomplish the BMR task of FBMC-OQAM signals in ICRNet nodes.A novel FBMC BMR algorithm is proposed with the transform channel convolution network(TCCNet)rather than a complicated two-dimensional convolution.Firstly,this is achieved by designing a low-complexity binary constellation diagram(BCD)gridding matrix as the input of TCCNet.Then,a transform channel convolution strategy is developed to convert the image-like BCD matrix into a serieslike data format,accelerating the BMR process while keeping discriminative features.Monte Carlo experimental results demonstrate that the proposed TCCNet obtains a performance gain of 8%and 40%over the traditional inphase/quadrature(I/Q)-based and constellation diagram(CD)-based methods at a signal noise ratio(SNR)of 12 dB,respectively.Moreover,the proposed TCCNet can achieve around 29.682 and 2.356 times faster than existing CD-Alex Network(CD-AlexNet)and I/Q-Convolutional Long Deep Neural Network(I/Q-CLDNN)algorithms,respectively.
文摘This paper proposes a novel open set recognition method,the Spatial Distribution Feature Extraction Network(SDFEN),to address the problem of electromagnetic signal recognition in an open environment.The spatial distribution feature extraction layer in SDFEN replaces convolutional output neural networks with the spatial distribution features that focus more on inter-sample information by incorporating class center vectors.The designed hybrid loss function considers both intra-class distance and inter-class distance,thereby enhancing the similarity among samples of the same class and increasing the dissimilarity between samples of different classes during training.Consequently,this method allows unknown classes to occupy a larger space in the feature space.This reduces the possibility of overlap with known class samples and makes the boundaries between known and unknown samples more distinct.Additionally,the feature comparator threshold can be used to reject unknown samples.For signal open set recognition,seven methods,including the proposed method,are applied to two kinds of electromagnetic signal data:modulation signal and real-world emitter.The experimental results demonstrate that the proposed method outperforms the other six methods overall in a simulated open environment.Specifically,compared to the state-of-the-art Openmax method,the novel method achieves up to 8.87%and 5.25%higher micro-F-measures,respectively.
基金This research was supported by the National Natural Science Foundation of China under Grant No.U19B2016.,and Zhejiang Provincial Key Lab of Data Storage and Transmission Technology,Hangzhou Dianzi University.
文摘Interference signals recognition plays an important role in anti-jamming communication.With the development of deep learning,many supervised interference signals recognition algorithms based on deep learning have emerged recently and show better performance than traditional recognition algorithms.However,there is no unsupervised interference signals recognition algorithm at present.In this paper,an unsupervised interference signals recognition method called double phases and double dimensions contrastive clustering(DDCC)is proposed.Specifically,in the first phase,four data augmentation strategies for interference signals are used in data-augmentation-based(DA-based)contrastive learning.In the second phase,the original dataset’s k-nearest neighbor set(KNNset)is designed in double dimensions contrastive learning.In addition,a dynamic entropy parameter strategy is proposed.The simulation experiments of 9 types of interference signals show that random cropping is the best one of the four data augmentation strategies;the feature dimensional contrastive learning in the second phase can improve the clustering purity;the dynamic entropy parameter strategy can improve the stability of DDCC effectively.The unsupervised interference signals recognition results of DDCC and five other deep clustering algorithms show that the clustering performance of DDCC is superior to other algorithms.In particular,the clustering purity of our method is above 92%,SCAN’s is 81%,and the other three methods’are below 71%when jammingnoise-ratio(JNR)is−5 dB.In addition,our method is close to the supervised learning algorithm.
基金supported by the National Natural Science Foundation of China(No.61771154)the Fundamental Research Funds for the Central Universities,China(No.3072021CF0815)supported by the Key Laboratory of Advanced Marine Communication and Information Technology,Ministry of Industry and Information Technology,Harbin Engineering University,Harbin,China。
文摘In the past ten years, many high-quality datasets have been released to support the rapid development of deep learning in the fields of computer vision, voice, and natural language processing. Nowadays, deep learning has become a key research component of the Sixth-Generation wireless systems(6G) with numerous regulatory and defense applications. In order to facilitate the application of deep learning in radio signal recognition, in this work, a large-scale real-world radio signal dataset is created based on a special aeronautical monitoring system-Automatic Dependent Surveillance-Broadcast(ADS-B). This paper makes two main contributions. First, an automatic data collection and labeling system is designed to capture over-the-air ADS-B signals in the open and real-world scenario without human participation. Through data cleaning and sorting, a high-quality dataset of ADS-B signals is created for radio signal recognition. Second, we conduct an in-depth study on the performance of deep learning models using the new dataset, as well as comparison with a recognition benchmark using machine learning and deep learning methods.Finally, we conclude this paper with a discussion of open problems in this area.
基金The authors wish to express their gratitude to the anonymous reviewers and the associate editor for their rigorous comments during the review process. In addition, authors also would like to thank SUN Chengbin and TAN Lei in our laboratory for their great contributions to the data-collection work. This work was supported by the National Natural Science Foundation of China (Grant Nos. 61571014 and 61601006), Beijing Nature Science Foundation (Grant No. 4172017), and Beijing Municipal Science and Technology Project (Grant No. Z161100001016003).
文摘The optical fiber pre-waming system (OFPS) has been gradually considered as one of the important means for pipeline safety monitoring. Intrusion signal types are correctly identified which could reduce the cost of troubleshooting and maintenance of the pipeline. Most of the previous feature extraction methods in OFPS are usually quested from the view of time domain. However, in some cases, there is no distinguishing feature in the time domain. In the paper, firstly, the intrusion signal features of the running, digging, and pick mattock are extracted in the frequency domain by multi-level wavelet decomposition, that is, the intrusion signals are decomposed into five bands. Secondly, the average energy ratio of different frequency bands is obtained, which is considered as the feature of each intrusion type. Finally, the feature samples are sent into the random vector functional-link (RVFL) network for training to complete the classification and identification of the signals. Experimental results show that the algorithm can correctly distinguish the different intrusion signals and achieve higher recognition rate.
文摘In this paper,we investigate the problem of key radar signal sorting and recognition in electronic intelligence(ELINT).Our major contribution is the development of a combined approach based on clustering and pulse repetition interval(PRI)transform algorithm,to solve the problem that the traditional methods based on pulse description word(PDW)were not exclusively targeted at tiny particular signals and were less time-efficient.We achieve this in three steps:firstly,PDW presorting is carried out by the DBSCAN(Density-Based Spatial Clustering of Applications with Noise)clustering algorithm,and then PRI estimates of each cluster are obtained by the PRI transform algorithm.Finally,by judging the matching between various PRI estimates and key targets,it is determined whether the current signal contains key target signals or not.Simulation results show that the proposed method should improve the time efficiency of key signal recognition and deal with the complex signal environment with noise interference and overlapping signals.
基金jointly funded by the Special Fund of Earthquake Industry(201508008)National Natural Science Foundation of China(41474051)Scientific Institution Innovation and Development Fund of Xinjiang Uygur Autonomous Region(201316)
文摘We used data from the Xinjiang Digital Seismic Network and PSD( Power Spectral Density) method to perform noise level assessment for six stations. We calculated the median of the power spectral density to evaluate the noise level of different stations. After the comparison of the power spectral density of different stations and the airgun signal recognition,we found that noise level of stations with recognizable airgun source signals is lower than that of stations without recognizable signals. The largest difference of the power spectral density is 40 d B,and the smallest is 15 d B. Finally,we found that the failure to recognize the signal of airgun sources at some stations is due to the noise level.
文摘In this Letter, a method based on the effects of imperfect oscillators in lasers is proposed to distinguish targets in continuous wave tracking lidar. This technique is based on the fact that each lidar signal source has a specific influence on the phase noise that makes real targets from the false ones. A simulated signal is produced by complex circuits, modulators, memory, and signal oscillators. For example, a deception laser beam has an unequal and variable phase noise from a real target. Thus, the phase noise of transmitted and received signals does not have the same power levels and patterns. To consider the performance of the suggested method, the probability of detection(PD) is shown for various signal-to-noise ratios and signal-to-jammer ratios based on experimental outcomes.