Recognition Module Research Articles

Low probability of interception radar overlapping signal modulation recognition based on an improved you-only-look-once version 8 network

Low probability of interception (LPI) radar is widely used in modern electronic warfare. With the increased radiation sources, multiple signals will arrive simultaneously. The traditional feature extraction method has too many features, which brings great trouble to the subsequent data processing. Most modulation recognition methods based on deep learning only consider the single signal after preprocessing, and the generalization ability is weak. This paper proposes a deep learning solution based on an improved you only look once version 8 (YOLOv8) network with a global attention mechanism (GAM), achieving recognition accuracy over 98% in a -10 dB signal-to-noise ratio (SNR) scenario to address the above problems. We improved the performance of the original network, which can be used in electronic countermeasures.

Acoustic modulation signal recognition based on endpoint detection

The modulation recognition technology for acoustic signals holds significant research importance in signal demodulation and communication signal reconnaissance, serving as a crucial component and key aspect. This paper investigates the modulation recognition technology for acoustic signals (< 20 kHz) from the perspectives of signal preprocessing and feature extraction. Firstly, it selects seven modulation signals 2ASK, 4ASK, 2FSK, 4FSK, 2PSK, 4PSK, and OFDM as recognition targets and systematically compares the effectiveness of four different endpoint detection algorithms in modulation signal recognition. To further enhance the performance of the short-time energy entropy ratio algorithm, this study introduces three different noise reduction algorithms for optimization. Finally, to accurately identify and distinguish between 2 and 4FSK signals, this study optimizes the related algorithms of the cyclic spectrum by using the kurtosis coefficient value Kur of the cyclic spectrum parameter matrix when the cyclic frequency α = 0 to differentiate between these two signals. The results show that at SNR of 4 dB, the proposed modulation recognition algorithm can effectively distinguish between these two signals, achieving a recognition accuracy of over 99%.

Intra-Pulse Modulation Recognition of Radar Signals Based on Efficient Cross-Scale Aware Network.

Radar signal intra-pulse modulation recognition can be addressed with convolutional neural networks (CNNs) and time-frequency images (TFIs). However, current CNNs have high computational complexity and do not perform well in low-signal-to-noise ratio (SNR) scenarios. In this paper, we propose a lightweight CNN known as the cross-scale aware network (CSANet) to recognize intra-pulse modulation based on three types of TFIs. The cross-scale aware (CSA) module, designed as a residual and parallel architecture, comprises a depthwise dilated convolution group (DDConv Group), a cross-channel interaction (CCI) mechanism, and spatial information focus (SIF). DDConv Group produces multiple-scale features with a dynamic receptive field, CCI fuses the features and mitigates noise in multiple channels, and SIF is aware of the cross-scale details of TFI structures. Furthermore, we develop a novel time-frequency fusion (TFF) feature based on three types of TFIs by employing image preprocessing techniques, i.e., adaptive binarization, morphological processing, and feature fusion. Experiments demonstrate that CSANet achieves higher accuracy with our TFF compared to other TFIs. Meanwhile, CSANet outperforms cutting-edge networks across twelve radar signal datasets, providing an efficient solution for high-precision recognition in low-SNR scenarios.

Ligation initiated self-priming isothermal polymerization enabled nano-signal amplification for APE1 sensing and logical unlocking

In this work, we developed a label-free and one-tube method for apurinic/apyrimidinic endonuclease 1 (APE1) sensing based on ligation initiated self-priming isothermal polymerization (SPIP) and signal amplification of fluorescent copper nanoparticles (CuNPs). This method was rationally proposed with modular design. In target recognition module, APE1 could specifically cleave substrate to release a linker DNA that served the ligation of an adenine/thymine (AT)-rich hairpin and a phosphorothioate-modified hairpin with the assistance of ligase. Then in SPIP amplification and signal transduction modules, once driven by polymerase, enzymatic extension and self-folding alternately occurred on the primary DNA originated from the ligation reaction, which resulted in accumulation of numerous AT-rich templates repeatedly included in the elongated dsDNA products, and ultimately enabled rapid formation of fluorescent CuNPs to output amplified signal. The accurate target identification and efficient signal amplification facilitated the highly selective and sensitive detection of APE1 with limit of detection of 8.6 × 10−4 U/mL. Benefiting from the robustness of this method, the normal cells and tumor cells could be distinguished unambiguously with this method according to the test results of cellular APE1. Moreover, the sensing strategy was employed in the concatenated AND logical operation and molecular keypad unlocking, which further displayed its extensibility and versatility.

Frequency Modulation Recognition for Radar Signal based on Resnet-LSTM Network

In order to satisfy the modulation recognition requirements for radar signal with large scale changes in pulse-width and bandwidth, the Resnet-LSTM network is proposed in this paper. First, the waveform is preprocessed by self-correlation, and then the I/Q waveform and self-correlation result are as the network input. Secondly, the local characteristics were extracted by Resnet network; and lastly the LSTM network is adopted to extract the serialization representation. The proposed algorithm avoids the flaw of CNN network cannot adapt to variable length, and problem of memory forgetting of the LSTM network for long sequence is reduced. The simulation shows that the modulation recognition accuracy is more than 96.1% over the detection sensitivity for LFM, Parabolic-NLFM, Triangular-NLFM, Sine-NLFM, Sawtooth-NLFM and FSK signals.

Radar Target Classification Using Enhanced Doppler Spectrograms with ResNet34_CA in Ubiquitous Radar

Ubiquitous Radar has become an essential tool for preventing bird strikes at airports, where accurate target classification is of paramount importance. The working mode of Ubiquitous Radar, which operates in track-then-identify (TTI) mode, provides both tracking information and Doppler information for the classification and recognition module. Moreover, the main features of the target’s Doppler information are concentrated around the Doppler main spectrum. This study innovatively used tracking information to generate a feature enhancement layer that can indicate the area where the main spectrum is located and combines it with the RGB three-channel Doppler spectrogram to form an RGBA four-channel Doppler spectrogram. Compared with the RGB three-channel Doppler spectrogram, this method increases the classification accuracy for four types of targets (ships, birds, flapping birds, and bird flocks) from 93.13% to 97.13%, an improvement of 4%. On this basis, this study integrated the coordinate attention (CA) module into the building block of the 34-layer residual network (ResNet34), forming ResNet34_CA. This integration enables the network to focus more on the main spectrum information of the target, thereby further improving the classification accuracy from 97.13% to 97.22%.

SSL-LRN: A Lightweight Semi-Supervised-Learning-Based Approach for UWA Modulation Recognition

Due to the lack of sufficient valid labeled data and severe channel fading, the recognition of various underwater acoustic (UWA) communication modulation types still faces significant challenges. In this paper, we propose a lightweight UWA communication type recognition network based on semi-supervised learning, named the SSL-LRN. In the SSL-LRN, a mean teacher–student mechanism is developed to improve learning performance by averaging the weights of multiple models, thereby improving recognition accuracy for insufficiently labeled data. The SSL-LRN employs techniques such as quantization and small convolutional kernels to reduce floating-point operations (FLOPs), enabling its deployment on underwater mobile nodes. To mitigate the performance loss caused by quantization, the SSL-LRN adopts a channel expansion module to optimize the neuron distribution. It also employs an attention mechanism to enhance the recognition robustness for frequency-selective-fading channels. Pool and lake experiments demonstrate that the framework effectively recognizes most modulation types, achieving a more than 5% increase in recognition accuracy at a 0 dB signal-to-noise ratio (SNRs) while reducing FLOPs by 84.9% compared with baseline algorithms. Even with only 10% labeled data, the performance of the SSL-LRN approaches that of the fully supervised LRN algorithm.

One2ThreeNet: An Automatic Microscale-Based Modulation Recognition Method for Underwater Acoustic Communication Systems

One2ThreeNet: An Automatic Microscale-Based Modulation Recognition Method for Underwater Acoustic Communication Systems

Low‐Interception Waveforms: To Prevent the Recognition of Spectrum Waveform Modulation via Adversarial Examples

AbstractDeep learning is applied to many complex tasks in the field of wireless communication, such as modulation recognition of spectrum waveforms, because of its convenience and efficiency. This leads to the problem of a malicious third party using a deep learning model to easily recognize the modulation format of the transmitted waveform. Some existing works address this problem directly using the concept of adversarial examples in the computer vision field without fully considering the characteristics of the waveform transmission in the physical world. Therefore, we propose two low‐interception waveforms (LIWs) generation methods, the LIW and ULIW algorithms, which can reduce the probability of the modulation being recognized by a third party without affecting the reliable communication of the friendly party. Among them, ULIW improves LIW algorithm by simulating channel noise during training cycle, and substantially reduces the perturbation magnitude while maintaining low interception accuracy. Our LIW and ULIW exhibit significant low‐interception performance in both numerical simulations and hardware experiments.

Modulation recognition for underwater acoustic communication based on hybrid neural network and feature fusion

It is a huge challenge for underwater acoustic receivers to correctly identify modulation methods due to the complex underwater channel environment and severe noise interference. Combined with the lightweight network (SqueezeNet) and attention mechanism (SENet), a multi-attribute and multi-scale feature fusion model based on a hybrid neural network is proposed, which achieves efficient and accurate recognition for modulation modes. First, the wavelet time-frequency (WTF) spectrum, square power spectrum, and contour maps of cyclic spectrum are extracted as multi-attribute inputs for the network to reduce the impact of inherent defects in single attribute feature. Second, shallow and deep features based on the SqueezeNet model are obtained as multi-scale features, of which the key feature expression ability is enhanced by the SENet model to provide sufficient feature information for modulation recognition. The simulation experiments and sea trial data confirm that the suggested method demonstrates strong generalization capabilities and effectiveness when applied to underwater acoustic channels and environmental noise. In contrast to algorithms in existence, the method verifies superior recognition abilities.

SiSe: Simultaneous and Sequential Transformers for multi-label activity recognition

Multi-label activity recognition is extremely challenging, where multiple activities may appear simultaneously or sequentially in a video. While previous works have realized the temporal co-occurrence of activities, the sequential order of activities have been largely overlooked. However, we argue that the sequential order of activities should also be preserved in correlation modeling, because shuffling the order might not form a semantically meaningful video. In this work, we present plug-and-play Simultaneous and Sequential Transformer (SiSe) modules for multi-label activity recognition. Upon frame features of all time steps, SiSe enhances spatio-temporal feature learning for multi-label activity recognition, by capturing the simultaneous and sequential activity correlations. Specifically, we employ a Simultaneous Transformer module to connect multiple activities that probably appear at each frame, and a hierarchical Sequential Transformer module to efficiently capture the sequential activity correlations in an order-preserved manner. Despite the straightforward and class-agnostic design of SiSe, it can outperform state-of-the-art approaches on three multi-label activity recognition benchmarks. In particular, we verify the significance of preserving the sequential order of activities with our Sequential Transformer in correlation modeling. We also conduct ablation studies and visual analysis for better understanding of our SiSe.

Enhancing radio signal classification under low SNR conditions using deep residual networks with channel attention mechanisms

This paper delves into the advancement of deep residual networks (ResNets) integrated with channel attention mechanisms for the classification of radio signals under conditions of low Signal-to-Noise Ratio (SNR). Utilizing an expansive dataset of radio signals, this paper introduces a novel architecture, MyResNet1, that combines residual learning with channel-wise attention, allowing the model to concentrate on essential features for precise classification. My, investigations exhibit notable improvements in classification accuracy, especially in challenging low SNR scenarios, highlighting the potential of attention-augmented deep residual networks in radio signal processing. Furthermore, this studyexplores various optimization strategies, including data augmentation and regularization techniques, to enhance the models performance and robustness. My findings contribute significantly to cognitive radio technologies and illuminate the potential of deep learning in sophisticated signal classification tasks, aligned with recent explorations in automatic modulation recognition (AMR) through deep learning and autoencoder-based methodologies for enhancing I/Q channel interactions.

Radio Signal Modulation Recognition Method Based on Hybrid Feature and Ensemble Learning: For Radar and Jamming Signals.

The detection performance of radar is significantly impaired by active jamming and mutual interference from other radars. This paper proposes a radio signal modulation recognition method to accurately recognize these signals, which helps in the jamming cancellation decisions. Based on the ensemble learning stacking algorithm improved by meta-feature enhancement, the proposed method adopts random forests, K-nearest neighbors, and Gaussian naive Bayes as the base-learners, with logistic regression serving as the meta-learner. It takes the multi-domain features of signals as input, which include time-domain features including fuzzy entropy, slope entropy, and Hjorth parameters; frequency-domain features, including spectral entropy; and fractal-domain features, including fractal dimension. The simulation experiment, including seven common signal types of radar and active jamming, was performed for the effectiveness validation and performance evaluation. Results proved the proposed method's performance superiority to other classification methods, as well as its ability to meet the requirements of low signal-to-noise ratio and few-shot learning.

Hook-Like DNAzyme-Activated Autocatalytic Biosensor for the Universal Detection of Pathogenic Bacteria.

DNAzyme-based assays have found extensive utility in pathogenic bacteria detection but often suffer from limited sensitivity and specificity. The integration of a signal amplification strategy could address this challenge, while the existing combination methods require extensive modification to accommodate various DNAzymes, limiting the wide-spectrum bacteria detection. We introduced a novel hook-like DNAzyme-activated autocatalytic nucleic acid circuit for universal pathogenic bacteria detection. The hook-like connector DNA was employed to seamlessly integrate the recognition element DNAzyme with the isothermal enzyme-free autocatalytic hybridization chain reaction and catalytic hairpin assembly for robust exponential signal amplification. This innovative autocatalytic circuit substantially amplifies the output signals from the DNAzyme recognition module, effectively overcoming DNAzyme's inherent sensitivity constraints in pathogen identification. The biosensor exhibits a strong linear response within a range of 1.5 × 103 to 3.7 × 107 CFU/mL, achieving a detection limit of 1.3 × 103 CFU/mL. Noted that the sensor's adaptability as a universal detection platform is established by simply modifying the hook-like connector module, enabling the detection of various pathogenic bacteria of considerable public health importance reported by the World Health Organization, including Escherichia coli, Staphylococcus aureus, Klebsiella pneumoniae, and Salmonella typhimurium. Additionally, the specificity of DNAzyme in bacterial detection is markedly improved due to the signal amplification process of the autocatalytic circuit. This hook-like DNAzyme-activated autocatalytic platform presents a versatile, sensitive, and specific approach for pathogenic bacteria detection, promising to significantly expand the applications of DNAzyme in bacteria detection.

Image-Acceleration Multimodal Danger Detection Model on Mobile Phone for Phone Addicts.

With the popularity of smartphones, a large number of "phubbers" have emerged who are engrossed in their phones regardless of the situation. In response to the potential dangers that phubbers face while traveling, this paper proposes a multimodal danger perception network model and early warning system for phubbers, designed for mobile devices. This proposed model consists of surrounding environment feature extraction, user behavior feature extraction, and multimodal feature fusion and recognition modules. The environmental feature module utilizes MobileNet as the backbone network to extract environmental description features from the rear-view image of the mobile phone. The behavior feature module uses acceleration time series as observation data, maps the acceleration observation data to a two-dimensional image space through GADFs (Gramian Angular Difference Fields), and extracts behavior description features through MobileNet, while utilizing statistical feature vectors to enhance the representation capability of behavioral features. Finally, in the recognition module, the environmental and behavioral characteristics are fused to output the type of hazardous state. Experiments indicate that the accuracy of the proposed model surpasses existing methods, and it possesses the advantages of compact model size (28.36 Mb) and fast execution speed (0.08 s), making it more suitable for deployment on mobile devices. Moreover, the developed image-acceleration multimodal phubber hazard recognition network combines the behavior of mobile phone users with surrounding environmental information, effectively identifying potential hazards for phubbers.

Rapid analysis of salbutamol residues in animal-derived food based on the fluorescence of MOFs/MIPs paper-based chips

Mimetic recognition materials for rapid and selective analysis of foodborne contamination from complex foodstuff becomes a hot research topic in the field of food safety. In this work, borate-affinity-functionalized MOF/molecularly imprinted polymer paper-based microfluidic chips (FSU-BA@MIP) were prepared by growing functionalized MOFs on paper surface with 3-carboxyphenylboronic acid (3-CPBA) as the ligands, followed by coating of a MIP layer of salbutamol (SAL) through a borate-affinity surface imprinting strategy. Using the obtained paper-based chip as a mimetic recognition and specific enrichment module, the fluorescence sensing platform achieved visual rapid quantitative analysis of SAL with a linear range of 0.1 × 10−2–5.0 mg L−1. The limits of detection and quantification were calculated to be 0.095 and 0.290 μg kg−1, respectively, and the sample loading via lateral flow aid could be completed within 30 min. The obtained MOFs/MIPs paper chips provide an effective way for the rapid analysis of harmful contaminants from different animal-derived foods.

Modulation recognition of underwater acoustic communication signals based on neural architecture search

Modulation recognition of underwater communication signals is a critical aspect of underwater information confrontation. However, the current deep learning-based methods for underwater communication modulation recognition often mimic neural network architectures used in image processing and speech processing, tending to perform poorly in low signal-to-noise ratio (SNR) conditions. This paper introduces a novel approach by utilizing neural architecture search (NAS) method. By automatically searching for neural architectures that are well-suited for modulation recognition in underwater environment, our proposed method improves the classification performance particularly in low SNR conditions. Furthermore, we have also proposed a recognition method based on attention mechanism and feature fusion, which substantially enhances the accuracy of identifying phase-modulated signals. Numerical simulation and experimental data are used to demonstrate performance of proposed methods.

The Rbfox1/LASR complex controls alternative pre-mRNA splicing by recognition of multi-part RNA regulatory modules.

The Rbfox proteins regulate alternative pre-mRNA splicing by binding to the RNA element GCAUG. In the nucleus, most of Rbfox is bound to LASR, a complex of RNA-binding proteins that recognize additional RNA motifs. However, it remains unclear how the different subunits of the Rbfox/LASR complex act together to bind RNA and regulate splicing. We used a nuclease-protection assay to map the transcriptome-wide footprints of Rbfox1/LASR on nascent cellular RNA. In addition to GCAUG, Rbfox1/LASR binds RNA containing motifs for LASR subunits hnRNPs M, H/F, C, and Matrin3. These elements are often arranged in tandem, forming multi-part modules of RNA motifs. To distinguish contact sites of Rbfox1 from the LASR subunits, we analyzed a mutant Rbfox1(F125A) that has lost RNA binding but remains associated with LASR. Rbfox1(F125A)/LASR complexes no longer interact with GCAUG but retain binding to RNA elements for LASR. Splicing analyses reveal that in addition to activating exons through adjacent GCAUG elements, Rbfox can also stimulate exons near binding sites for LASR subunits. Mini-gene experiments demonstrate that these diverse elements produce a combined regulatory effect on a target exon. These findings illuminate how a complex of RNA-binding proteins can decode combinatorial splicing regulatory signals by recognizing groups of tandem RNA elements.

Medical image intelligent diagnosis system based on facial emotion recognition and convolutional neural network

The facial emotional information of patients is one of the important indicators of their health status. By combining facial emotion recognition technology, medical imaging intelligent diagnostic systems can more comprehensively understand the psychological and physiological status of patients. This study first selected medical imaging datasets from the publicly available Medical Machine Learning Image Standard database and performed necessary preprocessing, including data cleaning, enhancement, and standardization. Next, a CNN model is constructed to extract key features from medical images and facial expression images through structures such as convolutional layers, activation functions, pooling layers, and fully connected layers. In addition, the system also integrates facial emotion recognition module and medical image analysis module, improving the comprehensiveness of diagnosis through feature fusion. Finally, the diagnostic results are displayed through a user interaction interface. Through comparative experiments, this study verified that the proposed CNN model outperforms traditional LSTM and SVM models in terms of accuracy, recall, and precision. The CNN model has shown significant advantages in medical imaging intelligent diagnosis systems, especially achieving a recall rate of 99%, demonstrating extremely high disease recognition ability. The results of this study not only provide new research directions for the field of intelligent diagnosis in medical imaging, but also lay a solid foundation for future clinical applications.

Medical image intelligent diagnosis system based on facial emotion recognition and convolutional neural network

The facial emotional information of patients is one of the important indicators of their health status. By combining facial emotion recognition technology, medical imaging intelligent diagnostic systems can more comprehensively understand the psychological and physiological status of patients. This study first selected medical imaging datasets from the publicly available Medical Machine Learning Image Standard database and performed necessary preprocessing, including data cleaning, enhancement, and standardization. Next, a CNN model is constructed to extract key features from medical images and facial expression images through structures such as convolutional layers, activation functions, pooling layers, and fully connected layers. In addition, the system also integrates facial emotion recognition module and medical image analysis module, improving the comprehensiveness of diagnosis through feature fusion. Finally, the diagnostic results are displayed through a user interaction interface. Through comparative experiments, this study verified that the proposed CNN model outperforms traditional LSTM and SVM models in terms of accuracy, recall, and precision. The CNN model has shown significant advantages in medical imaging intelligent diagnosis systems, especially achieving a recall rate of 99%, demonstrating extremely high disease recognition ability. The results of this study not only provide new research directions for the field of intelligent diagnosis in medical imaging, but also lay a solid foundation for future clinical applications.