Low SNR Environment Research Articles

Attention-Enhanced Dual-Branch Residual Network with Adaptive L-Softmax Loss for Specific Emitter Identification under Low-Signal-to-Noise Ratio Conditions

To address the issue associated with poor accuracy rates for specific emitter identification (SEI) under low signal-to-noise ratio (SNR) conditions, where the single-dimension radar signal characteristics are severely affected by noise, we propose an attention-enhanced dual-branch residual network structure based on the adaptive large-margin Softmax (ALS). Initially, we designed a dual-branch network structure to extract features from one-dimensional intermediate frequency data and two-dimensional time–frequency images, respectively. By assigning different attention weights according to their importance, these features are fused into an enhanced joint feature for further training. This approach enables the model to extract distinctive features across multiple dimensions and achieve good recognition performance even when the signal is affected by noise. In addition, we have introduced L-Softmax to replace the original Softmax and propose the ALS. This approach adaptively calculates the classification margin decision parameter based on the angle between samples and the classification boundary and adjusts the margin values of the sample classification boundaries; it reduces the intra-class distance for the same class while increasing the inter-class distance between different classes without the need for cumbersome experiments to determine the optimal value of decision parameters. Our experimental findings revealed that, in comparison to alternative methods, our proposed approach markedly enhances the model’s capability to extract features from signals and classify them in low-SNR environments, thereby effectively diminishing the influence of noise. Notably, it achieves the highest recognition rate across a range of low-SNR conditions, registering an average increase in recognition rate of 4.8%.

A Robust Super-resolution Algorithm in a Low SNR Environment for Vital Sign Radar

A Robust Super-resolution Algorithm in a Low SNR Environment for Vital Sign Radar

Robust Low-Cost Drone Detection and Classification Using Convolutional Neural Networks in Low SNR Environments

Robust Low-Cost Drone Detection and Classification Using Convolutional Neural Networks in Low SNR Environments

Automatic modulation recognition of radio fuzes using a DR2D-based adaptive denoising method and textural feature extraction

The identification of intercepted radio fuze modulation types is a prerequisite for decision-making in interference systems. However, the electromagnetic environment of modern battlefields is complex, and the signal-to-noise ratio (SNR) of such environments is usually low, which makes it difficult to implement accurate recognition of radio fuzes. To solve the above problem, a radio fuze automatic modulation recognition (AMR) method for low-SNR environments is proposed. First, an adaptive denoising algorithm based on data rearrangement and the two-dimensional (2D) fast Fourier transform (FFT) (DR2D) is used to reduce the noise of the intercepted radio fuze intermediate frequency (IF) signal. Then, the textural features of the denoised IF signal rearranged data matrix are extracted from the statistical indicator vectors of gray-level cooccurrence matrices (GLCMs), and support vector machines (SVMs) are used for classification. The DR2D-based adaptive denoising algorithm achieves an average correlation coefficient of more than 0.76 for ten fuze types under SNRs of −10 dB and above, which is higher than that of other typical algorithms. The trained SVM classification model achieves an average recognition accuracy of more than 96% on seven modulation types and recognition accuracies of more than 94% on each modulation type under SNRs of -12 dB and above, which represents a good AMR performance of radio fuzes under low SNRs.

Improving the Robustness of Time Difference of Arrival Estimation Based on the Energy Center of Gravity Rearrangement.

An accurate estimation of the time difference of arrival (TDOA) is crucial in localization, communication, and navigation. However, a low signal-to-noise ratio (SNR) can decrease the reliability of the TDOA estimation result. Therefore, this study aims to improve the performance of the TDOA estimation of dual-channel sensors for single-sound sources in low-SNR environments. This study introduces the theory of time rearrangement synchrosqueezing transform (TRST) into the time difference of arrival estimation. While the background noise TF points show random time delays, the signal time-frequency (TF) points originating from uniform directions that exhibit identical lags are considered in this study. In addition, the time difference rearrangement synchrosqueezing transform (TDST) algorithm is developed to separate the signal from the background noise by exploiting its distinct time delay characteristics. The implementation process of the proposed algorithm includes four main steps. First, a rough estimation of the time delay is performed by calculating the partial derivative of the short-time cross-power spectrum. Second, a rearrangement operation is conducted to separate the TF points of the signal and noise. Third, the TF points on both sides of the time-delay energy ridge are extracted. Finally, a refined TDOA estimation is realized by applying the inverse Fourier transformation on the extracted TF points. Furthermore, a second-order-based time difference reassigned synchrosqueezing transform algorithm is proposed to improve the robustness of the TDOA estimation by enhancing the TF energy aggregation. The proposed algorithms are verified by simulations and experiments. The results show that the proposed algorithms are more robust and accurate than the existing algorithms.

GGCNN: An Efficiency-Maximizing Gated Graph Convolutional Neural Network Architecture for Automatic Modulation Identification

Automatic modulation identification (AMI) is a technique to detect the modulation type and order of a received signal, which has the potential to enhance cognitive radio capabilities for future generations of communication devices. However, AMI classifiers traditionally have exhibited low efficiency in low signal-to-noise ratio (SNR) environments. Hence, to address this problem we present our novel Gated Graph Convolutional Neural Network (GGCNN) classifier for feature-based AMI. This architecture includes a robust feature extraction stage to extract deep correlative patterns about the received symbols. Not only does this feature extraction stage use the temporal characteristics of the received symbols, but it also takes advantage of embedded signaling features from the received signal. In the proposed classifier, the received constellations are treated as a graph, allowing it to outperform state-of-the-art classifiers due to its strong performance in graph classification. This is observed clearly in the visualization of the extracted features, even for high-order modulation schemes. In this paper, we present our systematic research conducted for maximizing the efficiency obtainable by our classifier. Extensive simulation results demonstrate a significant accuracy improvement of 18.44 percentage points, and an efficiency increase by 60.78% for our GGCNN-AMI classifier compared to state-of-the-art classifiers in low-SNR environments.

An Excess Kurtosis People Counting System Based on 1DCNN-LSTM Using Impulse Radio Ultra-Wide Band Radar Signals

As the Artificial Intelligence of Things (AIOT) and ubiquitous sensing technologies have been leaping forward, numerous scholars have placed a greater focus on the use of Impulse Radio Ultra-Wide Band (IR-UWB) radar signals for Region of Interest (ROI) population estimation. To address the problem concerning the fact that existing algorithms or models cannot accurately detect the number of people counted in ROI from low signal-to-noise ratio (SNR) received signals, an effective 1DCNN-LSTM model was proposed in this study to accurately detect the number of targets even in low-SNR environments with considerable people. First, human-induced excess kurtosis was detected by setting a threshold using the optimized CLEAN algorithm. Next, the preprocessed IR-UWB radar signal pulses were bundled into frames, and the resulting peaks were grouped to develop feature vectors. Subsequently, the sample set was trained based on the 1DCNN-LSTM algorithm neural network structure. In this study, the IR-UWB radar signal data were acquired from different real environments with different numbers of subjects (0–10). As indicated by the experimental results, the average accuracy of the proposed 1DCNN-LSTM model for the recognition of people counting reached 86.66% at ROI. In general, a high-accuracy, low-complexity, and high-robustness solution in IR-UWB radar people counting was presented in this study.

The Combination of Spectrum Subtraction and Cross-power Spectrum Phase Method for Time Delay Estimation

In order to solve the problem of large error of delay estimation in low SNR environment, a new delay estimation method based on cross power spectral frequency domain weighting and spectrum subtraction is proposed. Through theoretical analysis and MATLAB simulation, among the four common weighting functions, it is proved that the cross-power spectral phase weighting method has a good sharpening effect on the peak value of the cross-correlation function, and it is verified that the improved spectral subtraction method generally has a good noise reduction effect under different SNR environments. Finally, the joint simulation results of the whole algorithm show that the combination of spectrum subtraction and crosspower spectrum phase method can effectively sharpen the peak value of cross-correlation function and improve the accuracy of time delay estimation in the low SNR environment. The results of this paper can provide useful help for sound source localization in complex environments.

Hybrid feature extraction method of MFCC+GFCC helicopter noise based on wavelet decomposition

Aiming at the issue that the recognition accuracy of traditional acoustic signal features is low for helicopter acoustic signals with wind noise in the near field, a method of extracting mixed noise features of MFCC+GFCC based on wavelet decomposition is proposed. Firstly, the three-layer wavelet decomposition and reconstruction are applied to the helicopter acoustic signals; then, the Mel-Frequency Cepstral Coefficients (MFCC) and Gammatone-Frequency Cepstrum Coefficient (GFCC) are respectively extracted for the approximation and detail components; next, the coefficients of detail components which are averaged are combined with those of approximation components to form the hybrid feature parameters; finally, the convolutional neural network is used to classify the signal, to realize the correct recognition of helicopter acoustic signals. Experimental results show that the recognition accuracy is improved by almost 40% in contrast with other traditional methods, such as MFCC and GFCC, when the SNR is equal to -5dB. Further, When the SNR is -10dB, the recognition accuracy is more than 49%, while the traditional methods cannot effectively recognize the helicopter acoustic targets. The proposed feature extraction method can significantly improve the recognition accuracy in the low SNR environment, and provide a reference for near-field detection and recognition of helicopter acoustic targets.

Research on a Super-Resolution and Low-Complexity Positioning Algorithm Using FMCW Radar Based on OMP and FFT in 2D Driving Scene.

Multitarget positioning technology, such as FMCW millimeter-wave radar, has broad application prospects in autonomous driving and related mobile scenarios. However, it is difficult for existing correlation algorithms to balance high resolution and low complexity, and it is also difficult to ensure the robustness of the positioning algorithm using an aging antenna. This paper proposes a super-resolution and low-complexity positioning algorithm based on the orthogonal matching pursuit algorithm that can achieve more accurate distance and angle estimation for multiple objects in a low-SNR environment. The algorithm proposed in this paper improves the resolving power by two and one orders of magnitude, respectively, compared to the classical FFT and MUSIC algorithms in the same signal-to-noise environment, and the complexity of the algorithm can be reduced by about 25-30%, with the same resolving power as the OMP algorithm. Based on the positioning algorithm proposed in our paper, we use the PSO algorithm to optimize the arrangement of an aging antenna array so that its angle estimation accuracy is equivalent to that observed when the antenna is intact, improving the positioning algorithm's robustness. This paper also further realizes the use of the proposed algorithm and a single-frame intermediate frequency signal to estimate the position angle information of the object and obtain its motion trajectory and velocity, verifying the proposed algorithm's estimation ability when it comes to these qualities in a moving scene. Furthermore, this paper designs and carries out simulations and experiments. The experimental results verify that the positioning algorithm proposed in this paper can achieve accuracy, robustness, and real-time performance in autonomous driving scenarios.

Convolutional Deep Neural Network and Full Connectivity for Speech Enhancement

The speech signal that is received in real-time has background noise and reverberations, which have an impact on the quality of speech. Therefore, it is crucial to reduce or eliminate the noise and increase the intelligibility and quality of speech signals. In this study, a proposed method that is the most effective and challenging in a low SNR environment for three types of noise are removed, including washing machine, traffic noise, and electric fan noise, and clean speech is recovered. with three samples of noise which are mixed and added to the clean speech signal with a lower level of SNR value fixed at (-5, 0, 5) dBs, that noise source takes equal weights. The enhancement of the corrupted speech signal is done by applying a fully connected and convolutional neural network-based denoising algorithm and comparing their performance. The proposed network shows that a fully connected network (FCN) has less elapsed time than a convolutional network (CNN) while still achieving better performance, demonstrating its applicability for an embedded system. Also, the results obtained show that, overall, the CNN is better than the FCN regarding maximum coloration, PSNR, MES, and STOI.

Wavelet-Based Weighted Low-Rank Sparse Decomposition Model for Speech Enhancement Using Gammatone Filter Bank Under Low SNR Conditions

Estimating noise-related parameters in unsupervised speech enhancement (SE) techniques is challenging in low SNR and non-stationary noise environments. In the recent SE approaches, the best results are achieved by partitioning noisy speech spectrograms into low-rank noise and sparse speech parts. However, a few limitations reduce the performance of these SE methods due to the use of overlap and add in STFT process, noisy phase, due to inaccurate estimation of low rank in nuclear norm minimization and Euclidian distance measure in the cost function. These aspects can cause a loss of information in the reconstructed signal when compared to clean speech. To solve this, we propose a novel wavelet-based weighted low-rank sparse decomposition model for enhancing speech by incorporating a gammatone filter bank and Kullback–Leibler divergence. The proposed framework differs from other strategies in which the SE is carried entirely in time domain without the need for noise estimation. Further, to reduce the word error rate, these algorithms were trained and tested on a typical automatic speech recognition module. The experimental findings indicate that the proposed cascaded model has shown significant improvement under low SNR conditions over individual and traditional methods with regard to SDR, PESQ, STOI, SIG, BAK and OVL.

Automatic Modulation Recognition of Radiation Source Signals Based on Data Rearrangement and the 2D FFT

It is a challenge for automatic modulation recognition (AMR) methods for radiation source signals to work in environments with low signal-to-noise ratios (SNRs). This paper proposes a modulation feature extraction method based on data rearrangement and the 2D fast Fourier transform (FFT) (DR2D), and a DenseNet feature extraction network with early fusion is constructed to recognize the extracted modulation features. First, the input signal is preprocessed by DR2D to obtain three types of joint frequency feature bins with multiple time scales. Second, the feature fusion operation is performed on the inputs of the different layers of the proposed network. Finally, feature recognition is completed in the subsequent layers. The theoretical analysis and simulation results show that DR2D is a fast and robust preprocessing method for extracting the features of modulated radiation source signals with less computational complexity. The proposed DenseNet feature extraction network with early fusion can identify the extracted modulation features with less spatial complexity than other types of convolutional neural networks (CNNs) and performs well in low-SNR environments.

Multi-label Deep Forest: Towards Automatic Modulation Recognition of Compound Wireless Signals at Low-SNR Environment

Multi-label Deep Forest: Towards Automatic Modulation Recognition of Compound Wireless Signals at Low-SNR Environment

TSN-A: An Efficient Deep Learning Model for Automatic Modulation Classification Based on Intra-Class Confusion Reduction of Modulation Families

Automatic modulation classification (AMC) is an impressive technology, which is widely used in military and civilian fields. Recently, deep learning-based AMC (DL-AMC) methods have shown outstanding performance. However, most DL-AMC methods employ a single-stage network framework, which is effective for reducing only inter-class confusion of modulation families, while the intra-class confusion of modulation families attracts less attention and remains to be addressed. This letter proposes a two-stage network with attention (TSN-A), which consists of a main network and two sub-networks. In the first stage, the main network can effectively tackle the problem of inter-class confusion. In the second stage, the sub-networks can efficiently reduce the intra-class confusion of AM-SSB and AM-DSB families. TSN-A achieves the top results on the public challenging RML 2018.01a dataset and is more competitive in low SNR environments. Meanwhile, our main network exhibits stronger and more robust classification ability on different datasets than the benchmark models. Moreover, our sub-networks are very applicable to be migrated to other DL-AMC models for helping them improve the classification accuracy by reducing the intra-class confusion.

Low SNR Multi-Emitter Signal Sorting and Recognition Method Based on Low-Order Cyclic Statistics CWD Time-Frequency Images and the YOLOv5 Deep Learning Model.

It is difficult for traditional signal-recognition methods to effectively classify and identify multiple emitter signals in a low SNR environment. This paper proposes a multi-emitter signal-feature-sorting and recognition method based on low-order cyclic statistics CWD time-frequency images and the YOLOv5 deep network model, which can quickly dissociate, label, and sort the multi-emitter signal features in the time-frequency domain under a low SNR environment. First, the denoised signal is extracted based on the low-order cyclic statistics of the typical modulation types of radiation source signals. Second, the time-frequency graph of multisource signals was obtained through CWD time-frequency analysis. The cyclic frequency was controlled to balance the noise suppression effect and operation time to achieve noise suppression of multisource signals at a low SNR. Finally, the YOLOv5s deep network model is used as a classifier to sort and identify the received signals from multiple radiation sources. The method proposed in this paper has high real-time performance. It can identify the radiation source signals of different modulation types with high accuracy under the condition of a low SNR.

TSUNet: Transformer-based single channel speech enhancement using SkipConv U-Net for cochlear implant (CI) users

The presence of multi-speaker babble noise greatly degrades speech intelligibility for communications with cochlear implant (CI) users. Convolutional Neural Network (CNN) based speech enhancement has been popular for suppressing noise because it can localize spectro-temporal information within the feature set. However, suppressing noise without creating artifacts is challenging in low SNR environments, and even more so if the noise is speech-like such as Babble noise. Transformers have emerged as a useful architecture with innate global self-attention mechanisms to capture long-range dependencies and global context, which is a limitation for CNN. The Mel spectrogram has a strong correlation along both time and frequency axes, and thus captures contextual information beneficial for speech enhancement. In this study, we propose TSUNet, which leverages both Transformers and U-Net to capture sufficient low-level details of contextual information in the speech time-frequency domain. TSUNet employs Transformer layers at the bottleneck of CNN based on U-Net. We also incorporate a neural vocoder to synthesize speech from the time-frequency representation without using a contaminated phase. The performance of the proposed network is evaluated under simulated and real recordings of noisy speech including CI testing. The proposed network achieves very effective performance in both scenarios.

Polar Code Parameter Recognition Algorithm Based on Dual Space

To improve the performance of polar code parameter recognition in the fields of intelligent communication, communication detection, and network countermeasures, we propose a new recognition scheme for the additive white Gaussian noise (AWGN) channel. The scheme turns parameter recognition problems into hypothetical tests and is effective due to the check relationship between the received codeword and the dual space determined by the correct parameters. First, a sub-matrix is obtained by removing the frozen-bit-index rows of the polar code generator matrix, and then its dual matrix is calculated. To check the relationship of the dual matrix and codewords, the average likelihood ratio of codewords is introduced as a test statistic, and then the corresponding decision threshold is deduced. Next, the degree of conformity of polar code recognition is defined, and the minimum code length and code rate corresponding to the highest degree of conformity are chosen to calculate the index of information-bit positions by a Gaussian approximation (GA) construction algorithm. Finally, the chosen code length, code rate, and corresponding index are provided as the recognition results. Simulation results show that the algorithm can achieve effective parameter recognition in both high-signal-to-noise (SNR) and low-SNR environments, and that the algorithm’s recognition performance increases with decreasing code rate and code length. The parameter recognition rate with a code length of 128 and code rate of 1/5 is close to 100% when the SNR is 4 dB, and the algorithm complexity increases almost linearly with the decreasing code rate and the increasing length of the intercepted data.

Adaptive Diagonal Loading Technique to Improve Direction of Arrival Estimation Accuracy for Linear Antenna Array Sensors

Diagonal loading is one of the most widely used and effective methods to improve the robustness of both adaptive beamformers and Direction of Arrival (DOA) estimation due to the involvement of the sensor received covariance matrix. In addition, subspace-based DOA estimation techniques rely on multiple snapshots to achieve high estimation accuracies. This paper presents the study of a modified diagonally loaded sample covariance matrix for accurate DOA estimation in adverse scenarios. The proposed and novel technique deciphers poor DOA estimation in a low SNR environment by computationally changing the received sample covariance matrix. Our method is computationally simple as it does not require peak searching and does not depend on the coherency of the signal. The efficacy of the proposed method is examined via computer simulation for various sensor array sizes and the number of snapshot samples. Based on our numerical simulation results, our proposed method generally outperforms most state-of-the-art DOA estimators. In a finite number of snapshots and a single signal source, our proposed method performs 9.5% better than the state-of-the-art DOA estimation technique, 2.8% in multiple signal sources, and 8.5% in a single snapshot, single signal source environment of gained DOA estimation performance.

Speech Enhancement for Robust Speech Recognition Using Weighted Low Rank and Sparse Decomposition Models under Low SNR Conditions

Noise estimation is a crucial stage in speech enhancement (SE), and it commonly necessitates the use of prior models for speech, noise, or both. Prior models, on the other hand, can be ineffective in dealing with unseen nonstationary noise, especially at low signal to noise (SNR) levels. This paper proposes to assess the efficacy of an unsupervised SE approach based on weighted low rank and sparse matrix factorization to estimate noise and speech when neither is available beforehand by decomposing the input noisy spectrum into a low-rank noise component and a sparse speech component. Due to the approximation of the actual rank of noise, these techniques are constrained, and they do not directly exploit the low-rank property in optimization. Nuclear norm minimization (NNM) is the most well-known approach, as it can precisely recover the matrix's rank under certain restricted and theoretical guarantee conditions. NNM, on the other hand, is unable to reliably estimate the matrix rank in many situations. Significant advancements in computer vision and machine learning applications have demonstrated that a weighted nuclear norm minimization (WNNM), overcomes NNM shortcomings, and achieves a superior matrix rank approximation than NNM. Consequently, in this study, we present alternate SE algorithms that make use of weighted low rank and sparsity constraints to separate speech and noise spectrograms. Following that, they were trained and evaluated on a standard Automatic Speech Recognition (ASR) engine to lower the Word Error Rate (WER). Extensive investigations on the impact of real-world noise on speech signals show that the proposed model outperforms the existing state of art models in terms of objective measures like SDR, PESQ, SIG, BAK, OVL, and STOI values in varied noise circumstances under low SNR environments.