Physical Communication最新文献_第5页

Shortwave signal modulation recognition method using adaptive time-Frequency threshold denoising and feature fusion 短波信号调制识别方法采用自适应时频阈值去噪和特征融合

IF 2.2 4区计算机科学 Q3 ENGINEERING, ELECTRICAL & ELECTRONIC

Physical Communication

Pub Date : 2026-03-01 Epub Date: 2026-01-06 DOI: 10.1016/j.phycom.2026.102994

Chen Shen , Tingting Lyu , Yu Li , Tianqi Lin , Yulong Liu

Automatic modulation classification (AMC) techniques are crucial for cognitive radio and communication systems. However, in low signal-to-noise ratio (SNR) conditions, transient shortwave signals are highly vulnerable to noise interference. This vulnerability leads to a reduction in identification accuracy. Medium time scale shortwave signals offer more stable characteristics. However, these signals are influenced by the time-varying SNR. This effect causes the energy density distribution to become discrete, thereby leading to lower recognition accuracy. To address this issue, this paper proposes a new architecture combining the adaptive time-frequency threshold denoising (ATFTD) algorithm and dual-modal feature fusion to enhance the modulation recognition accuracy of medium time scale shortwave signals. First, the signals are transformed into two types of time-frequency images (TFIs) using smoothed pseudo Wigner-Ville distribution (SPWVD) and Born-Jordan distribution (BJD). Subsequently, the ATFTD algorithm denoises these two TFIs. Next, the denoised TFIs are input into deep networks for feature extraction, and Jensen-Shannon divergence (JSD) is employed for fusion. Meanwhile, the time-domain statistical features of the signals are extracted and concatenated with the fused TFI features. Finally, the concatenated features are fed into a fully connected network for classification. Experimental results demonstrate that the proposed solution achieves over 90% recognition accuracy across six deep learning networks (AlexNet, ResNet18, VGGNet16, DenseNet121, ResNet50, and ResNet152), with the best performance observed in the ResNet152 network, ultimately reaching an average recognition accuracy of 99.625%.

自动调制分类（AMC）技术是认知无线电通信系统的关键技术。然而，在低信噪比条件下，瞬态短波信号极易受到噪声干扰。这个漏洞会降低识别的准确性。中时间尺度短波信号具有更稳定的特性。然而，这些信号受到时变信噪比的影响。这种影响导致能量密度分布变得离散，从而导致识别精度降低。针对这一问题，本文提出了一种将自适应时频阈值去噪（ATFTD）算法与双峰特征融合相结合的新架构，以提高中时间尺度短波信号的调制识别精度。首先，利用平滑伪Wigner-Ville分布（SPWVD）和Born-Jordan分布（BJD）将信号变换成两种时频图像（tfi）。随后，ATFTD算法对这两个tfi进行去噪。然后，将去噪后的tfi输入深度网络进行特征提取，并利用Jensen-Shannon散度（JSD）进行融合。同时，提取信号的时域统计特征，并与融合后的TFI特征进行拼接。最后，将连接的特征输入到一个全连接的网络中进行分类。实验结果表明，该方案在六个深度学习网络（AlexNet、ResNet18、VGGNet16、DenseNet121、ResNet50和ResNet152）上的识别准确率超过90%，其中ResNet152网络的识别准确率最高，达到99.625%的平均识别准确率。

{"title":"Shortwave signal modulation recognition method using adaptive time-Frequency threshold denoising and feature fusion","authors":"Chen Shen , Tingting Lyu , Yu Li , Tianqi Lin , Yulong Liu","doi":"10.1016/j.phycom.2026.102994","DOIUrl":"10.1016/j.phycom.2026.102994","url":null,"abstract":"<div><div>Automatic modulation classification (AMC) techniques are crucial for cognitive radio and communication systems. However, in low signal-to-noise ratio (SNR) conditions, transient shortwave signals are highly vulnerable to noise interference. This vulnerability leads to a reduction in identification accuracy. Medium time scale shortwave signals offer more stable characteristics. However, these signals are influenced by the time-varying SNR. This effect causes the energy density distribution to become discrete, thereby leading to lower recognition accuracy. To address this issue, this paper proposes a new architecture combining the adaptive time-frequency threshold denoising (ATFTD) algorithm and dual-modal feature fusion to enhance the modulation recognition accuracy of medium time scale shortwave signals. First, the signals are transformed into two types of time-frequency images (TFIs) using smoothed pseudo Wigner-Ville distribution (SPWVD) and Born-Jordan distribution (BJD). Subsequently, the ATFTD algorithm denoises these two TFIs. Next, the denoised TFIs are input into deep networks for feature extraction, and Jensen-Shannon divergence (JSD) is employed for fusion. Meanwhile, the time-domain statistical features of the signals are extracted and concatenated with the fused TFI features. Finally, the concatenated features are fed into a fully connected network for classification. Experimental results demonstrate that the proposed solution achieves over 90% recognition accuracy across six deep learning networks (AlexNet, ResNet18, VGGNet16, DenseNet121, ResNet50, and ResNet152), with the best performance observed in the ResNet152 network, ultimately reaching an average recognition accuracy of 99.625%.</div></div>","PeriodicalId":48707,"journal":{"name":"Physical Communication","volume":"75 ","pages":"Article 102994"},"PeriodicalIF":2.2,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145980718","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Adaptive transmit antenna grouping scheme for signed quadrature spatial modulation 符号正交空间调制的自适应发射天线分组方案

IF 2.2 4区计算机科学 Q3 ENGINEERING, ELECTRICAL & ELECTRONIC

Physical Communication

Pub Date : 2026-03-01 Epub Date: 2026-02-20 DOI: 10.1016/j.phycom.2026.103049

Yu Guan , Lexi Xu , Yuyang Peng , Enjian Bai , Xue Qin Jiang , Han Hai

Recently, signed quadrature spatial modulation (SQSM) has been proposed to improve the spectral efficiency (SE) for spatial modulation systems. SQSM extends the real and imaginary dimensions along with their inverse counterparts to transmit more bits. However, the number of transmit antennas of SQSM is limited to a power of 2. In this paper, a novel transmit antenna grouping scheme for SQSM (AG-SQSM) and an adaptive version for AG-SQSM (AAG-SQSM) are proposed. The AG-SQSM scheme makes full use of space resources, breaking through the limitation of SQSM. The SE of AG-SQSM is higher than that of traditional modulations when the number of transmit antennas increases linearly. Moreover, AAG-SQSM improves the bit error rate (BER) of the system by adopting an adaptive algorithm. The BER upper bounds of AG-SQSM and AAG-SQSM over Rayleigh fading channels are analyzed. Simulation results reveal that AG-SQSM and AAG-SQSM both have better BER performance than conventional SQSM when using the maximum likelihood (ML) detection algorithm.

近年来，为了提高空间调制系统的频谱效率，提出了符号正交空间调制（SQSM）。SQSM扩展实维和虚维以及它们的逆维来传输更多的比特。然而，SQSM的发射天线数量被限制为2的幂。本文提出了一种新的面向SQSM的发射天线分组方案（AG-SQSM）和一种面向AG-SQSM的自适应方案（AAG-SQSM）。AG-SQSM方案充分利用了空间资源，突破了SQSM的局限性。当发射天线数量线性增加时，AG-SQSM的SE高于传统调制。此外，AAG-SQSM采用自适应算法提高了系统的误码率。分析了AG-SQSM和AAG-SQSM在瑞利衰落信道下的误码率上界。仿真结果表明，当使用最大似然（ML）检测算法时，AG-SQSM和AAG-SQSM都比传统SQSM具有更好的误码率性能。

引用次数: 0

Energy harvesting-assisted rate-splitting multiple access under external and internal overhearing: Secrecy analysis 外部和内部监听下能量收集辅助分频多址：保密性分析

IF 2.2 4区计算机科学 Q3 ENGINEERING, ELECTRICAL & ELECTRONIC

Physical Communication

Pub Date : 2026-03-01 Epub Date: 2026-02-21 DOI: 10.1016/j.phycom.2026.103056

Khuong Ho-Van

The study investigates the security susceptibilities of rate-splitting multiple access (RSMA) systems powered by harvested energy, focusing on threats from both external and internal eavesdropping. External eavesdropping arises from unauthorized wiretappers unrelated to RSMA, whereas internal eavesdropping is an inherent issue of successive interference cancellation, wherein any RSMA receiver may intercept others’ signals. This study also evaluates the security capability of energy harvesting-assisted RSMA considering realistic energy harvesting conditions. The proposed evaluation framework is confirmed through Monte Carlo simulation. The findings reveal that, across various operational scenarios, RSMA provides superior security compared to orthogonal multiple access and non-orthogonal multiple access in energy harvesting wireless systems.

该研究调查了由收集能量供电的分频多址（RSMA）系统的安全脆弱性，重点关注来自外部和内部窃听的威胁。外部窃听来自与RSMA无关的未经授权的窃听者，而内部窃听是一个固有的连续干扰消除问题，其中任何一个RSMA接收者都可能拦截他人的信号。考虑实际能量收集条件，评估了能量收集辅助RSMA的安全能力。通过蒙特卡罗仿真验证了所提出的评价框架。研究结果表明，在各种操作场景中，与能量收集无线系统中的正交多址和非正交多址相比，RSMA提供了更高的安全性。

引用次数: 0

Enhancing communication performance with DFT-S-OFDM and multi-dimensional index modulation 利用DFT-S-OFDM和多维指数调制技术提高通信性能

IF 2.2 4区计算机科学 Q3 ENGINEERING, ELECTRICAL & ELECTRONIC

Physical Communication

Pub Date : 2026-03-01 Epub Date: 2026-02-05 DOI: 10.1016/j.phycom.2026.103027

Yongzhi Yu , Hao Li , Ge Li , Limin Guo

The Discrete Fourier Transform Spread Orthogonal Frequency Division Multiplexing (DFT-S-OFDM) system provides low Peak Average Power Ratio (PAPR) and low complexity. However, the current mode of subcarrier mapping in the system leads to a waste of subcarriers, thereby reducing the transmission rate. In this paper, a DFT-S-OFDM with Multi-Dimensional Index (DFT-S-OFDM-MDI) modulation system is proposed. The MDI incorporates subcarrier index (SI) and constellation diagram index (CDI), enabling the transmission of bit information in diverse index modes, thereby enhancing the system's transmission efficiency. We have configured six constellation diagram schemes for the system and incorporated spreading matrices to significantly boost the Bit Error Rate (BER) performance. The simulation results indicate that the proposed system is superior to traditional methods in terms of Spectral Efficiency (SE), Energy Efficiency (EE), and Bit Error Rate (BER).

离散傅立叶变换扩展正交频分复用（DFT-S-OFDM）系统具有低峰值平均功率比（PAPR）和低复杂度的特点。然而，目前系统中的子载波映射模式导致了子载波的浪费，从而降低了传输速率。本文提出了一种具有多维指数的DFT-S-OFDM （DFT-S-OFDM- mdi）调制系统。MDI结合了子载波索引（SI）和星座图索引（CDI），能够以多种索引方式传输比特信息，从而提高了系统的传输效率。我们为系统配置了六种星座图方案，并加入了扩频矩阵，以显著提高误码率（BER）性能。仿真结果表明，该系统在频谱效率（SE）、能量效率（EE）和误码率（BER）方面都优于传统方法。

引用次数: 0

Toward dynamic radar signal sorting via gramian dimensionality elevation fusion CNN 基于格兰曼维度高程融合CNN的雷达信号动态分选

IF 2.2 4区计算机科学 Q3 ENGINEERING, ELECTRICAL & ELECTRONIC

Physical Communication

Pub Date : 2026-03-01 Epub Date: 2026-02-26 DOI: 10.1016/j.phycom.2026.103057

Hongyuan Liu , Xingjun Lai , Tengfei Li , Zichuan Yu , Dayi Lin , Deyan Kong , Mingming Chen

Radar signal sorting (RSS) is a crucial component within electronic reconnaissance systems, aiming to separate multiple radar pulses from an interlaced pulse stream. However, with the increasing complexity of the electromagnetic environment, radar signals are exhibiting diversified and dynamic characteristics. Traditional RSS techniques struggle to handle the nonlinear variations inherent in complex modulated signals and are susceptible to noise interference and spurious pulses. To address these challenges, deep learning technologies, notably Convolutional Neural Networks (CNNs), offer a promising solution for sorting signals with dynamic parameters. Nevertheless, single-modality CNNs lack the capacity to simultaneously capture intricate feature representations and process sequential data effectively. Motivated by this limitation, this paper proposes a novel network named Gramian-based CNN Fusion Network (GCF-Net) designed for efficient RSS in dynamic environments. Initially, comprehensive sorting datasets are established by modeling seven types of Pulse Repetition Interval (PRI) sequences. Subsequently, a dual-branch recognition and classification network is developed based on the Gramian Angular Difference Field (GADF) representation. Finally, experiments are conducted using the proposed GCF-Net. Experimental results demonstrate that GCF-Net achieves a high recognition rate of 92.4%. Moreover, it exhibits exceptional robustness in highly dynamic environments and significantly outperforms other deep-learning-based RSS methods.

雷达信号分选（RSS）是电子侦察系统中的重要组成部分，旨在从交错脉冲流中分离多个雷达脉冲。然而，随着电磁环境的日益复杂，雷达信号呈现出多样化和动态性的特点。传统的RSS技术难以处理复杂调制信号中固有的非线性变化，并且容易受到噪声干扰和杂散脉冲的影响。为了应对这些挑战，深度学习技术，特别是卷积神经网络（cnn），为分类具有动态参数的信号提供了一个很有前途的解决方案。然而，单模态cnn缺乏同时捕获复杂特征表示和有效处理序列数据的能力。基于这一局限性，本文提出了一种基于gramian的CNN融合网络（GCF-Net），旨在实现动态环境下的高效RSS。首先，通过对7种脉冲重复间隔（PRI）序列进行建模，建立了综合排序数据集。在此基础上，建立了基于格拉曼角差场（GADF）表示的双分支识别分类网络。最后，利用本文提出的GCF-Net进行了实验。实验结果表明，GCF-Net的识别率达到了92.4%。此外，它在高动态环境中表现出出色的鲁棒性，显著优于其他基于深度学习的RSS方法。

{"title":"Toward dynamic radar signal sorting via gramian dimensionality elevation fusion CNN","authors":"Hongyuan Liu , Xingjun Lai , Tengfei Li , Zichuan Yu , Dayi Lin , Deyan Kong , Mingming Chen","doi":"10.1016/j.phycom.2026.103057","DOIUrl":"10.1016/j.phycom.2026.103057","url":null,"abstract":"<div><div>Radar signal sorting (RSS) is a crucial component within electronic reconnaissance systems, aiming to separate multiple radar pulses from an interlaced pulse stream. However, with the increasing complexity of the electromagnetic environment, radar signals are exhibiting diversified and dynamic characteristics. Traditional RSS techniques struggle to handle the nonlinear variations inherent in complex modulated signals and are susceptible to noise interference and spurious pulses. To address these challenges, deep learning technologies, notably Convolutional Neural Networks (CNNs), offer a promising solution for sorting signals with dynamic parameters. Nevertheless, single-modality CNNs lack the capacity to simultaneously capture intricate feature representations and process sequential data effectively. Motivated by this limitation, this paper proposes a novel network named Gramian-based CNN Fusion Network (GCF-Net) designed for efficient RSS in dynamic environments. Initially, comprehensive sorting datasets are established by modeling seven types of Pulse Repetition Interval (PRI) sequences. Subsequently, a dual-branch recognition and classification network is developed based on the Gramian Angular Difference Field (GADF) representation. Finally, experiments are conducted using the proposed GCF-Net. Experimental results demonstrate that GCF-Net achieves a high recognition rate of 92.4%. Moreover, it exhibits exceptional robustness in highly dynamic environments and significantly outperforms other deep-learning-based RSS methods.</div></div>","PeriodicalId":48707,"journal":{"name":"Physical Communication","volume":"75 ","pages":"Article 103057"},"PeriodicalIF":2.2,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147397594","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

TDMRNet: Exploiting multiscale residual learning for blind estimation of polar code parameters TDMRNet：利用多尺度残差学习盲估计极性码参数

IF 2.2 4区计算机科学 Q3 ENGINEERING, ELECTRICAL & ELECTRONIC

Physical Communication

Pub Date : 2026-03-01 Epub Date: 2026-01-29 DOI: 10.1016/j.phycom.2026.103028

Chang Li , Yanou Cao , Kai Liu

Polar codes are widely utilized in modern communication systems, where accurate identification of their parameters from received signals is crucial for reliable information decoding. However, existing methods often perform inadequately in non-cooperative scenarios due to the lack of prior information. In this study, we propose a two-dimensional multiscale residual network (TDMRNet) for the blind identification of polar code parameters. Considering the structural characteristics of polar codes, the received signal sequence is first reshaped into a two-dimensional matrix to enable more effective parameter recognition. To further process the reshaped data, TDMRNet integrates an inter-code feature extraction module (ICFEM) and a data-coupling module (DCM). The ICFEM employs a unit convolution layer and three parallel standard convolution layers to construct a multiscale feature extractor, thereby enhancing the model's generalization capability and recognition accuracy. The DCM combines standard and dimension-reducing residual blocks to strengthen feature extraction while mitigating overfitting. Simulation results demonstrate that TDMRNet achieves a recognition accuracy of 93.80 %, surpassing existing methods, and reducing computational complexity, confirming its effectiveness in scenarios without prior knowledge.

极性码在现代通信系统中得到广泛应用，从接收到的信号中准确识别极性码的参数对于可靠的信息解码至关重要。然而，由于缺乏先验信息，现有的方法在非合作情况下往往表现不佳。在这项研究中，我们提出了一种二维多尺度残差网络（TDMRNet）来盲识别极性码参数。考虑到极化码的结构特点，首先将接收到的信号序列重构为二维矩阵，从而实现更有效的参数识别。为了进一步处理重构数据，TDMRNet集成了代码间特征提取模块（ICFEM）和数据耦合模块（DCM）。ICFEM采用一个单元卷积层和三个并行的标准卷积层构建多尺度特征提取器，从而提高了模型的泛化能力和识别精度。DCM结合了标准残差块和降维残差块，在减少过拟合的同时加强了特征提取。仿真结果表明，TDMRNet的识别准确率达到93.80%，超越了现有方法，并且降低了计算复杂度，验证了其在无先验知识场景下的有效性。

{"title":"TDMRNet: Exploiting multiscale residual learning for blind estimation of polar code parameters","authors":"Chang Li , Yanou Cao , Kai Liu","doi":"10.1016/j.phycom.2026.103028","DOIUrl":"10.1016/j.phycom.2026.103028","url":null,"abstract":"<div><div>Polar codes are widely utilized in modern communication systems, where accurate identification of their parameters from received signals is crucial for reliable information decoding. However, existing methods often perform inadequately in non-cooperative scenarios due to the lack of prior information. In this study, we propose a two-dimensional multiscale residual network (TDMRNet) for the blind identification of polar code parameters. Considering the structural characteristics of polar codes, the received signal sequence is first reshaped into a two-dimensional matrix to enable more effective parameter recognition. To further process the reshaped data, TDMRNet integrates an inter-code feature extraction module (ICFEM) and a data-coupling module (DCM). The ICFEM employs a unit convolution layer and three parallel standard convolution layers to construct a multiscale feature extractor, thereby enhancing the model's generalization capability and recognition accuracy. The DCM combines standard and dimension-reducing residual blocks to strengthen feature extraction while mitigating overfitting. Simulation results demonstrate that TDMRNet achieves a recognition accuracy of 93.80 %, surpassing existing methods, and reducing computational complexity, confirming its effectiveness in scenarios without prior knowledge.</div></div>","PeriodicalId":48707,"journal":{"name":"Physical Communication","volume":"75 ","pages":"Article 103028"},"PeriodicalIF":2.2,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147397923","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Open set recognition for drone based on deep metric learning 基于深度度量学习的无人机开放集识别

IF 2.2 4区计算机科学 Q3 ENGINEERING, ELECTRICAL & ELECTRONIC

Physical Communication

Pub Date : 2026-03-01 Epub Date: 2026-02-04 DOI: 10.1016/j.phycom.2026.103030

Jiangfeng Hong, Jiakai Liang, Chao Wang, Keqiang Yue, Wenjun Li

With the rapid development of drone technology and its wide range of applications, some potential security issues have gradually surfaced. In practice, however, most drone identification methods are based on the closed set assumption, which hinders the capacity to respond effectively to unidentified drone classes.To address this issue, we propose an open set recognition method based on deep metric learning in this paper. Specifically, we design a multi-scale time-frequency feature extraction network that learns and extracts low-dimensional embedding representations from the Time-Frequency Spectrum (TFS) of Unmanned Aerial Vehicle (UAV) radio-frequency signals. In the subsequent feature optimization stage, we jointly employ triplet loss and center loss to refine the feature distribution. Moreover, we directly use the distances between the optimized class centers and the feature vectors as the basis for classification, leading to a simple and deployment-friendly prototype-distance based recognition rule that avoids introducing an additional trainable classifier module and does not require test-time iterative clustering to form prototypes. Finally, we propose an Adaptive Threshold Distance Distribution Classifier (AT-DDC), which dynamically determines the optimal decision threshold by analyzing the distance distribution between feature vectors and class centers, thus enabling open set recognition of drones.Compared with the existing method, the proposed method’s known average accuracy (KA) has been improved by 3.5% in all scenarios. Meanwhile, the true unknown rate (TUR) is reduced by only 1.40%, and the mean absolute percentage error (MAPE) between the decision thresholds and the optimal thresholds is reduced by 8.68%, which demonstrates the excellent performance of the proposed method in drone open set recognition.

随着无人机技术的快速发展和广泛应用，一些潜在的安全问题也逐渐浮出水面。然而，在实践中，大多数无人机识别方法都是基于闭集假设，这阻碍了有效响应未知无人机类别的能力。为了解决这一问题，本文提出了一种基于深度度量学习的开放集识别方法。具体而言，我们设计了一个多尺度时频特征提取网络，从无人机（UAV）射频信号的时频频谱（TFS）中学习和提取低维嵌入表示。在随后的特征优化阶段，我们联合使用三重损失和中心损失来细化特征分布。此外，我们直接使用优化的类中心与特征向量之间的距离作为分类的基础，从而产生了一个简单且易于部署的基于原型距离的识别规则，该规则避免了引入额外的可训练分类器模块，并且不需要测试时间迭代聚类来形成原型。最后，我们提出了一种自适应阈值距离分布分类器（AT-DDC），该分类器通过分析特征向量与类中心之间的距离分布动态确定最优决策阈值，从而实现无人机的开放集识别。与现有方法相比，该方法在所有场景下的已知平均精度（KA）提高了3.5%。同时，真实未知率（TUR）仅降低了1.40%，决策阈值与最优阈值之间的平均绝对百分比误差（MAPE）降低了8.68%，表明该方法在无人机开放集识别中具有优异的性能。

{"title":"Open set recognition for drone based on deep metric learning","authors":"Jiangfeng Hong, Jiakai Liang, Chao Wang, Keqiang Yue, Wenjun Li","doi":"10.1016/j.phycom.2026.103030","DOIUrl":"10.1016/j.phycom.2026.103030","url":null,"abstract":"<div><div>With the rapid development of drone technology and its wide range of applications, some potential security issues have gradually surfaced. In practice, however, most drone identification methods are based on the closed set assumption, which hinders the capacity to respond effectively to unidentified drone classes.To address this issue, we propose an open set recognition method based on deep metric learning in this paper. Specifically, we design a multi-scale time-frequency feature extraction network that learns and extracts low-dimensional embedding representations from the Time-Frequency Spectrum (TFS) of Unmanned Aerial Vehicle (UAV) radio-frequency signals. In the subsequent feature optimization stage, we jointly employ triplet loss and center loss to refine the feature distribution. Moreover, we directly use the distances between the optimized class centers and the feature vectors as the basis for classification, leading to a simple and deployment-friendly prototype-distance based recognition rule that avoids introducing an additional trainable classifier module and does not require test-time iterative clustering to form prototypes. Finally, we propose an Adaptive Threshold Distance Distribution Classifier (AT-DDC), which dynamically determines the optimal decision threshold by analyzing the distance distribution between feature vectors and class centers, thus enabling open set recognition of drones.Compared with the existing method, the proposed method’s known average accuracy (KA) has been improved by 3.5% in all scenarios. Meanwhile, the true unknown rate (TUR) is reduced by only 1.40%, and the mean absolute percentage error (MAPE) between the decision thresholds and the optimal thresholds is reduced by 8.68%, which demonstrates the excellent performance of the proposed method in drone open set recognition.</div></div>","PeriodicalId":48707,"journal":{"name":"Physical Communication","volume":"75 ","pages":"Article 103030"},"PeriodicalIF":2.2,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147397930","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Cell-free massive MIMO signal detector fused with multi-head attention by graph neural network 基于图神经网络融合的无小区海量MIMO信号检测器

IF 2.2 4区计算机科学 Q3 ENGINEERING, ELECTRICAL & ELECTRONIC

Physical Communication

Pub Date : 2026-03-01 Epub Date: 2026-01-29 DOI: 10.1016/j.phycom.2026.103023

Shihao Guo , Xiaohui Zhang , Gaoyuan Zhang

The Cell-free massive multiple-input multiple-output (CF-mMIMO) system utilizes multiple access points (APs) to receive user information and has the advantage of uniform coverage. During uplink transmission, the aliased multi-user signals are aggregated by each AP to the central processing unit (CPU). Designing efficient detection algorithms is the key to eliminating multi-user interference and ensuring collaborative gain. In this paper, aiming at the limitations of the independent Gaussian approximation in expectation propagation (EP), and using the 2D multi-head attention (MHA) module to obtain the temporal and frequency correlations, a Graph neural network (GNN) detector based on EP and MHA is proposed. It is called a GNN-MHA-EP detector. The simulation results show that in different low-precision quantization systems, or when there are errors in channel estimation, the proposed detection algorithm is significantly superior to the traditional EP algorithm, belief propagation (BP) algorithm and minimum mean square error (MMSE) algorithm, and also significantly improves the performance of the GNN-EP detection algorithm and maintains the same order of computational complexity with it.

无小区大规模多输入多输出（CF-mMIMO）系统利用多个接入点（ap）接收用户信息，具有均匀覆盖的优点。在上行链路传输过程中，混叠的多用户信号由每个AP汇聚到中央处理器（CPU）。设计高效的检测算法是消除多用户干扰、保证协同增益的关键。针对独立高斯近似在期望传播（EP）中的局限性，利用二维多头注意（MHA）模块获取期望传播的时间和频率相关性，提出了一种基于EP和MHA的图神经网络（GNN）检测器。它被称为GNN-MHA-EP探测器。仿真结果表明，在不同的低精度量化系统中，或信道估计存在误差时，所提出的检测算法明显优于传统的EP算法、信念传播（BP）算法和最小均方误差（MMSE）算法，并且显著提高了GNN-EP检测算法的性能，并与之保持相同的计算复杂度。

{"title":"Cell-free massive MIMO signal detector fused with multi-head attention by graph neural network","authors":"Shihao Guo , Xiaohui Zhang , Gaoyuan Zhang","doi":"10.1016/j.phycom.2026.103023","DOIUrl":"10.1016/j.phycom.2026.103023","url":null,"abstract":"<div><div>The Cell-free massive multiple-input multiple-output (CF-mMIMO) system utilizes multiple access points (APs) to receive user information and has the advantage of uniform coverage. During uplink transmission, the aliased multi-user signals are aggregated by each AP to the central processing unit (CPU). Designing efficient detection algorithms is the key to eliminating multi-user interference and ensuring collaborative gain. In this paper, aiming at the limitations of the independent Gaussian approximation in expectation propagation (EP), and using the 2D multi-head attention (MHA) module to obtain the temporal and frequency correlations, a Graph neural network (GNN) detector based on EP and MHA is proposed. It is called a GNN-MHA-EP detector. The simulation results show that in different low-precision quantization systems, or when there are errors in channel estimation, the proposed detection algorithm is significantly superior to the traditional EP algorithm, belief propagation (BP) algorithm and minimum mean square error (MMSE) algorithm, and also significantly improves the performance of the GNN-EP detection algorithm and maintains the same order of computational complexity with it.</div></div>","PeriodicalId":48707,"journal":{"name":"Physical Communication","volume":"75 ","pages":"Article 103023"},"PeriodicalIF":2.2,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147397953","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

FMCW radar implementation on RF sampling transceiver with signal processing techniques for enhanced range accuracy FMCW雷达在射频采样收发机上的实现，采用信号处理技术提高距离精度

IF 2.2 4区计算机科学 Q3 ENGINEERING, ELECTRICAL & ELECTRONIC

Physical Communication

Pub Date : 2026-03-01 Epub Date: 2026-01-20 DOI: 10.1016/j.phycom.2026.103013

S. Reshma , Indu Gopan , M.J. Lal , S. Sreelal , M. Vani Devi

In this work, we present the implementation of an improved signal processing algorithm on an FMCW radar system realized using a direct RF sampling transceiver with greater flexibility compared to traditional radars built with several independent analog and digital components. In addition to the sweep bandwidth, the range resolution of an FMCW radar relies on the beat frequency estimation technique used in the signal processing stage. For improved range accuracy, we propose a signal processing algorithm based on the Chirp Z transform (CZT) combined with an inter-bin interpolation technique that outperforms the conventional FFT in beat frequency estimation. To further enhance the accuracy of radar range, the wavelet denoising technique is applied to the beat signal prior to beat frequency estimation. The effectiveness of this algorithm is validated through hardware experiment in radiated mode using the FMCW signal impaired with simulated noisy scenarios, especially phase noise. The range estimation accuracy of the radar system was evaluated based on Root Mean Square Error (RMSE) as the performance metric. From the test results, it was found that the CZT technique with the Jacobsen estimator and wavelet denoising resulted in the least RMSE value and provided the best accuracy in range measurements, even under noisy conditions.

在这项工作中，我们提出了一种改进的信号处理算法在FMCW雷达系统上的实现，该系统使用直接射频采样收发器实现，与使用多个独立模拟和数字组件构建的传统雷达相比，该系统具有更大的灵活性。除了扫描带宽外，FMCW雷达的距离分辨率还依赖于信号处理阶段使用的拍频估计技术。为了提高距离精度，我们提出了一种基于Chirp Z变换（CZT）和帧间插值技术的信号处理算法，该算法在拍频估计方面优于传统的FFT。为了进一步提高雷达距离的精度，在拍频估计之前对拍频信号进行小波去噪处理。通过硬件实验验证了该算法在辐射模式下的有效性，实验中，FMCW信号受到模拟噪声特别是相位噪声的干扰。以均方根误差（RMSE）为性能指标，对雷达系统的距离估计精度进行了评价。从测试结果中可以发现，使用Jacobsen估计器和小波去噪的CZT技术即使在噪声条件下，也能产生最小的RMSE值，并提供最佳的距离测量精度。

{"title":"FMCW radar implementation on RF sampling transceiver with signal processing techniques for enhanced range accuracy","authors":"S. Reshma , Indu Gopan , M.J. Lal , S. Sreelal , M. Vani Devi","doi":"10.1016/j.phycom.2026.103013","DOIUrl":"10.1016/j.phycom.2026.103013","url":null,"abstract":"<div><div>In this work, we present the implementation of an improved signal processing algorithm on an FMCW radar system realized using a direct RF sampling transceiver with greater flexibility compared to traditional radars built with several independent analog and digital components. In addition to the sweep bandwidth, the range resolution of an FMCW radar relies on the beat frequency estimation technique used in the signal processing stage. For improved range accuracy, we propose a signal processing algorithm based on the Chirp Z transform (CZT) combined with an inter-bin interpolation technique that outperforms the conventional FFT in beat frequency estimation. To further enhance the accuracy of radar range, the wavelet denoising technique is applied to the beat signal prior to beat frequency estimation. The effectiveness of this algorithm is validated through hardware experiment in radiated mode using the FMCW signal impaired with simulated noisy scenarios, especially phase noise. The range estimation accuracy of the radar system was evaluated based on Root Mean Square Error (RMSE) as the performance metric. From the test results, it was found that the CZT technique with the Jacobsen estimator and wavelet denoising resulted in the least RMSE value and provided the best accuracy in range measurements, even under noisy conditions.</div></div>","PeriodicalId":48707,"journal":{"name":"Physical Communication","volume":"75 ","pages":"Article 103013"},"PeriodicalIF":2.2,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146024416","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Range estimation in FMCW SDR radars with unsynchronized receivers 非同步接收机FMCW SDR雷达的距离估计

IF 2.2 4区计算机科学 Q3 ENGINEERING, ELECTRICAL & ELECTRONIC

Physical Communication

Pub Date : 2026-03-01 Epub Date: 2026-01-05 DOI: 10.1016/j.phycom.2026.103000

Juan Carlos Martinez Quintero, Edward Paul Guillen Pinto

Frequency-modulated continuous wave (FMCW) radars are widely used in applications such as automotive sensing and remote monitoring. Implementing these systems on software-defined radio (SDR) platforms offers flexibility but also introduces challenges, such as synchronization and isolation between the transmitter and receiver, processing large volumes of data, and digital signal processing. This paper proposes two methods for range detection in FMCW radars using unsynchronized SDR platforms for the transmitter and receiver. In both approaches, a direct signal between the transmitting and receiving antennas, positioned a few centimeters apart, serves as a reference. The first method employs FFT on the squared magnitude of the received signal to estimate the range, while the second method adapts the classical matched filter technique. The proposed methods also enable the receiver to operate with a lower sampling rate than the transmitter. Tests were conducted using LimeSDR platforms with a frequency deviation of 58 MHz (corresponding to the sweep bandwidth), an output power of 10 dBm, and a central frequency of 2.4 GHz, targeting distances between 5.27 m and 28.45 m. The first method achieved an average error of 3.5% with matching sampling rates, while the second method reduced the error to 2.15%. When halving the receiver's sampling rate, the average errors increased to 4.28% and 2.33%, respectively. These results demonstrate the potential of the proposed methods for flexible and efficient radar implementations on SDR platforms.

调频连续波（FMCW）雷达广泛应用于汽车传感和远程监控等领域。在软件定义无线电（SDR）平台上实现这些系统提供了灵活性，但也带来了挑战，例如发射器和接收器之间的同步和隔离，处理大量数据和数字信号处理。本文提出了两种基于非同步SDR平台的FMCW雷达距离检测方法。在这两种方法中，发射天线和接收天线之间相隔几厘米的直接信号作为参考。第一种方法采用FFT对接收信号的幅值平方估计距离，第二种方法采用经典的匹配滤波技术。所提出的方法还使接收机能够以比发射机更低的采样率工作。测试使用LimeSDR平台进行，频率偏差为58 MHz（对应于扫描带宽），输出功率为10 dBm，中心频率为2.4 GHz，目标距离为5.27 m至28.45 m。在采样率匹配的情况下，第一种方法的平均误差为3.5%，第二种方法的平均误差为2.15%。当接收机采样率减半时，平均误差分别增加到4.28%和2.33%。这些结果证明了所提出的方法在SDR平台上灵活有效地实现雷达的潜力。

{"title":"Range estimation in FMCW SDR radars with unsynchronized receivers","authors":"Juan Carlos Martinez Quintero, Edward Paul Guillen Pinto","doi":"10.1016/j.phycom.2026.103000","DOIUrl":"10.1016/j.phycom.2026.103000","url":null,"abstract":"<div><div>Frequency-modulated continuous wave (FMCW) radars are widely used in applications such as automotive sensing and remote monitoring. Implementing these systems on software-defined radio (SDR) platforms offers flexibility but also introduces challenges, such as synchronization and isolation between the transmitter and receiver, processing large volumes of data, and digital signal processing. This paper proposes two methods for range detection in FMCW radars using unsynchronized SDR platforms for the transmitter and receiver. In both approaches, a direct signal between the transmitting and receiving antennas, positioned a few centimeters apart, serves as a reference. The first method employs FFT on the squared magnitude of the received signal to estimate the range, while the second method adapts the classical matched filter technique. The proposed methods also enable the receiver to operate with a lower sampling rate than the transmitter. Tests were conducted using LimeSDR platforms with a frequency deviation of 58 MHz (corresponding to the sweep bandwidth), an output power of 10 dBm, and a central frequency of 2.4 GHz, targeting distances between 5.27 m and 28.45 m. The first method achieved an average error of 3.5% with matching sampling rates, while the second method reduced the error to 2.15%. When halving the receiver's sampling rate, the average errors increased to 4.28% and 2.33%, respectively. These results demonstrate the potential of the proposed methods for flexible and efficient radar implementations on SDR platforms.</div></div>","PeriodicalId":48707,"journal":{"name":"Physical Communication","volume":"75 ","pages":"Article 103000"},"PeriodicalIF":2.2,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145915144","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0