A one-dimension convolutional neural network based interference classification method

Interference is a common problem in wireless communication, navigation and radar systems. A wide variety of interferences are used to degrade the communication quality especially in electronic warfare environment. In modern military communication systems, interference classification is an important module for its ability to obtain prior interference information before adopting related anti-interference method. This paper proposes a deep learning based interference classification method, which applies one-dimension convolutional neural networks to automatically extract interference features for classification. Computer simulations show better classification performance and lower computational complexity. Meanwhile, this proposed method is implied on software defined radios (SDR) hardware, more than 99% correct classification probability can be achieved with limited samples of the received signal, which verifies the robustness of this proposed method.     

Deep learning based physical layer security for terrestrial communications in 5G and beyond networks: A survey

The key principle of physical layer security (PLS) is to permit the secure transmission of confidential data using efficient signal-processing techniques. Also, deep learning (DL) has emerged as a viable option to address various security concerns and enhance the performance of conventional PLS techniques in wireless networks. DL is a strong data exploration technique which can be used to learn normal and abnormal behavior of 5G and beyond wireless networks in an insecure channel paradigm. Also, since DL techniques can successfully predict future new instances by learning from existing ones, they can successfully predict new attacks, which frequently involve mutations of earlier attacks. Thus, motivated by the benefits of DL and PLS, this survey provides a comprehensive review that overviews how DL-based PLS techniques can be employed for solving various security concerns in 5G and beyond networks. The survey begins with an overview of physical layer threats and security concerns in 5G and beyond networks. Then, we present a detailed analysis of various DL and deep reinforcement learning (DRL) techniques that are applicable to PLS applications. We present the specific use-cases of PLS design for each type of technique, including attack detection, physical layer authentication (PLA), and other PLS techniques. Then, we present an in-depth overview of the key areas of PLS where DL can be used to enhance the security of wireless networks, such as automatic modulation classification (AMC), secure beamforming, PLA, etc. Performance evaluation metrics for DL-based PLS design are subsequently covered. Finally, we provide insights to the readers about various challenges and future research trends in the design of DL-based PLS for terrestrial communications in 5G and beyond networks.     

Security concerns on machine learning solutions for 6G networks in mmWave beam prediction

6G – sixth generation – is the latest cellular technology currently under development for wireless communication systems. In recent years, machine learning (ML) algorithms have been applied widely in various fields, such as healthcare, transportation, energy, autonomous cars, and many more. Those algorithms have also been used in communication technologies to improve the system performance in terms of frequency spectrum usage, latency, and security. With the rapid developments of ML techniques, especially deep learning (DL), it is critical to consider the security concern when applying the algorithms. While ML algorithms offer significant advantages for 6G networks, security concerns on artificial intelligence (AI) models are typically ignored by the scientific community so far. However, security is also a vital part of AI algorithms because attackers can poison the AI model itself. This paper proposes a mitigation method for adversarial attacks against proposed 6G ML models for the millimeter-wave (mmWave) beam prediction using adversarial training. The main idea behind generating adversarial attacks against ML models is to produce faulty results by manipulating trained DL models for 6G applications for mmWave beam prediction. We also present a proposed adversarial learning mitigation method’s performance for 6G security in mmWave beam prediction application a fast gradient sign method attack. The results show that the defended model under attack’s mean square errors (i.e., the prediction accuracy) are very close to the undefended model without attack.     

Learning to short-time Fourier transform in spectrum sensing

The future wireless communication will come up with a strict requirement on high spectral efficiency, developing novel algorithms for spectrum sensing with deep sensing capability will be more challenging. However, traditional expert feature-based spectrum sensing algorithms are lack of sufficient capability of self-learning and adaptability to unknown environments and complex cognitive tasks. To address this problem, we propose to build up a deep learning network to learn short time-frequency transformation (STFT), a basic entity of traditional spectrum sensing algorithms. Spectrum sensing based on the learning to STFT network is supposed to automatically extract features for communication signals and makes decisions for complex cognitive tasks meanwhile. The feasibility and performances of the designed learning network are verified by classifying signal modulation types in deep spectrum sensing applications.     

Signal propagation techniques for wireless underground communication networks

Wireless Underground Communication Networks (WUCNs) consist of wireless devices that operate below the ground surface. These devices are either (i) buried completely under dense soil, or (ii) placed within a bounded open underground space, such as underground mines and road/subway tunnels. The main difference between WUCNs and the terrestrial wireless communication networks is the communication medium. In this paper, signal propagation characteristics are described in these constrained areas. First, a channel model is described for electromagnetic (EM) waves in soil medium. This model characterizes not only the propagation of EM waves, but also other effects such as multipath, soil composition, water content, and burial depth. Second, the magnetic induction (MI) techniques are analyzed for communication through soil. Based on the channel model, the MI waveguide technique for communication is developed to address the high attenuation challenges of MI waves through soil. Furthermore, a channel model, i.e., the multimode model, is provided to characterize the wireless channel for WUCNs in underground mines and road/subway tunnels. The multimode model can characterize two cases for underground communication, i.e., the tunnel channel and the room-and-pillar channel. Finally, research challenges for the design communication protocols for WUCNs in both underground environments are discussed based on the analysis of the signal propagation.     

Spectrum sensing in cognitive radio networks and metacognition for dynamic spectrum sharing between radar and communication system: A review

The massive growth in mobile users and wireless technologies has resulted in increased data traffic and created demand for additional radio spectrum. This growing demand for radio spectrum has resulted in spectrum congestion and mandated the need for coexistence between radar and interfering communication emitters. To address the aforementioned issues, it is critical to review existing policies and evaluate new technologies that can utilize spectrum in an efficient and intelligent manner. Cognitive radio and cognitive radar are two promising technologies that exploit spectrum using dynamic spectrum access techniques. Additionally, introducing the bio-inspired concept ‘metacognition’ in a cognitive process has shown to increase the effectiveness and robustness of the cognitive radio and cognitive radar system. Metacognition is a high-order thinking agent that monitors and regulates the cognition process through a feedback and control process called the perception–action cycle. Extensive research has been done in the field of spectrum sensing in cognitive radio and spectral coexistence between radar and communication systems. This paper provides a detailed classification of spectrum sensing schemes and explains how dynamic spectrum access strategies share the spectrum between radar and communication systems. In addition to this, the fundamentals of cognitive radio, its architecture, spectrum management framework, and metacognition concept in radar are discussed. Furthermore, this paper presents various research issues, challenges, and future research directions associated with spectrum sensing in cognitive radar and dynamic spectrum access strategies in cognitive radar.     

A Signal processing method of OFDM communication receiver based on CNN

Orthogonal frequency division multiplexing (OFDM) the signal processing is a key issue in wireless communication research. The multipath effect and Doppler shift of wireless communication channels can lead to distortion of the transmitted signal, which poses a considerable challenge to the information recovery of communication receivers. This paper presents the signal processing method of OFDM communication based on convolutional neural network (CNN). The method replaces all signal processing modules of the OFDM communication receiver with CNN, and the information is recovered by the CNN. In order to adapt to the processing of communication signals, we designed a one-dimensional convolutional neural network (1D-CONV-CNN) model as the neural network structures by this method. Simulation results indicate that the signal processing method effectively reduces the bit error rate (BER) and improves its performance compared with the conventional reception method under different channel conditions.     

Physical layer security using beamforming techniques for 5G and beyond networks: A systematic review

It has been predicted that by the year 2030, 5G and beyond 5G (B5G) networks are expected to provide hundreds of trillions of gigabytes of data for various emerging applications such as augmented, mixed, and virtual reality (AR/MR/VR), wireless computer-brain interfaces (WCBI), connected robotics and autonomous systems. Most of these applications share data with each other using an open channel, i.e., the Internet. The open and broadcast nature of wireless channel makes the communication susceptible to various types of attacks (e.g., eavesdropping, jamming). Thus, there is a strong requirement to enhance the secrecy of wireless channel to maintain the privacy and confidentiality of transmitted data. Physical layer security (PLS) has evolved as a novel concept and robust alternative to cryptography-based techniques, which have a number of drawbacks and practical issues for 5G and beyond networks. Beamforming is an energy-efficient PLS technique, that involves steering of the transmitted signal in a particular direction, while considering that an intruding user attempts to decode the transmitted data. Motivated from these points, this article summarizes various beamforming based PLS techniques for secure data transmission in 5G and B5G networks. We investigate the eight most promising techniques for beamforming in PLS: Non-Orthogonal Multiple Access (NOMA), Full-Duplex Networks, Massive Multiple-Input Multiple-Output (MIMO), Cognitive Radio (CR) Network, Relay Network, Simultaneous Wireless Information and Power Transfer (SWIPT), UAV Communication Networks and Space Information Networks, and Heterogeneous Networks. Moreover, various physical layer threats and countermeasures associated with 5G and B5G networks are subsequently covered. Lastly, we provide insights to the readers about constraints and challenges for the usage of beamforming-based PLS techniques in various upcoming future applications.     

Accurate modulation classification under impaired wireless channels via shallow convolutional neural networks

Classifying the modulation type of radio signals plays an important role in current and future wireless communication systems. We present a modulation classification method based on convolutional neural networks that reaches high accuracy in face of various channel characteristics and signal conditions without requiring the network to have a very large depth. Experiment results show that the proposed method reaches accurate classification under different system impairment settings that include sampling rate offset, carrier frequency offset, multi-path fading, and additive white Gaussian noise. For instance, compared to a state-of-the-art method, accuracy is improved up to 25% in classifying difficult modulation types under system impairments. Source code of the proposed method is available online.     

Large-Scale Location-Aware Services in Access: Hierarchical Building/Floor Classification and Location Estimation Using Wi-Fi Fingerprinting Based on Deep Neural Networks

We report the results of our investigation on the use of deep neural networks (DNNs) for building/floor classification and floor-level location estimation based on Wi-Fi fingerprinting. We propose a new DNN architecture based on a stacked autoencoder for feature space dimension reduction and a feed-forward classifier for multi-label classification with arg max functions to convert multi-label classification results into multi-class classification ones. We also demonstrate a prototype system for floor-level location estimation using received signal strengths measured on XJTLU campus. Our results show the strengths of DNN-based approaches, providing near state-of-the-art performance with less parameter tuning and higher scalability.     

Channel identification with Improved Variational Mode Decomposition

The identification of the type of wireless propagation channel (e.g., Line of Sight (LOS) or Non Line of Sight (NLOS)) is an important function in the wireless communication design and deployment especially in rich propagation environments. The wireless channel characteristics can be quite specific not only between Line of Sight (LOS) and Non Line of Sight (NLOS) wireless propagation conditions but also in different NLOS environments.In recent times, machine learning approaches have been increasingly used to differentiate and classify channel characteristics and this paper is part of this trend. In particular, this paper proposes the combination of machine learning with a recently proposed signal processing tool called Variational Mode Decomposition (VMD), which is a decomposition algorithm that decomposes a time series into several modes which have specific sparsity properties. VMD itself is a refinement of the Empirical Mode Decomposition (EMD) and demonstrated a superior performance to EMD for classification problems. One issue for the practical deployment of VMD in channel identification problems is the presence of hyper-parameters, which must be tuned for the applied context. The main contribution of this paper is to propose a novel approach for channel identification based on an improvement of VMD called Improved Variational Mode Decomposition (IVMD), where the optimal values of the hyper-parameters of VMD are automatically identified on the basis of the Shannon entropy of the signal output from the channel. Then, various features are extracted from the modes generated by IVMD and a sequential feature selection algorithm is applied to select the optimal features. This paper applies the proposed approach with IVMD to a data set generated by the authors with a wireless channel emulator, where 6 different propagation scenarios (including no fading conditions) are created for WiFi 802.11g signals, where only the preamble is used for channel identification. Even if channel identification based on the normalized preamble is a challenging classification problem, the proposed IVMD is able to outperform significantly the application of basic VMD, EMD and the time and frequency domain representations (as commonly done in literature) of the WiFi signals.     

Lightweight authentication scheme for massive MIMO on Internet of Things connectivity

Massive MIMO is an essential technology in developing 5G networks and a concept that may be applied to other wireless systems. However, the advantages of adopting massive MIMO for broadband communication are well-established. Recently researcher has been devoted to building communication systems sustaining high communication rates with security. While massive MIMO for Internet-of-Things (IoT) connectivity is still a developing issue, IoT connectivity has requirements and limitations that differ significantly from broadband connections. Although IoT makes people’s lives easier by allowing physical devices to flow through, the interaction of open wireless channels such as Bluetooth, ZigBee, LoRa, Narrowband-Internet of Things (NBIoT), and WiFi, has produced various security and privacy difficulties. Identity authentication is one of the effective solutions for addressing the Internet of Things security and privacy concerns. The typical point-to-point authentication technique ignores the internet of things’ massive number of nodes and limited node resources. Group authentication is an authentication method that simultaneously confirms the identity of a group of members, offering a novel approach to identity identification for the internet of things nodes. However, existing group authentication systems appropriate for the internet of things scenarios pose security issues. They cannot withstand malicious attacks such as forging and replay and cannot prevent group managers from fooling group members. Most existing group authentication schemes are computationally expensive and cannot be applied to resource-constrained IoT scenarios. At the same time, existing systems based on secret-sharing technology cannot resist forgery attacks and replay attacks. The attacker can forge a legal token by modifying the Lagrangian coefficient in the authentication token to pass the group authentication. This work employs verifiable secret sharing technology to create a lightweight verifiable group authentication method (L-IoT-GS) suitable for Internet of Things situations to resist group managers’ deceptive group behavior. Nodes in the Internet of Things scenario can frequently join and leave the network. Because of this, this article proposes a critical update link based on a verified group authentication system for updating group Member rights. According to security analysis, the suggested L-IoT-GS scheme meets accuracy and confidentiality requirements and can withstand malicious attacks such as replay, forgery, and impersonation. Furthermore, performance study and experimental simulation reveal that the L-IoT-GS technique minimizes group members’ computing costs compared to existing standard IoT group authentication schemes.     

基于无线传感器网络的地震信号特征提取方法研究

为解决野外古墓葬安防网络中高采样率会缩短无线传感器网络寿命的问题, 提出了使用功率谱二次处理对地震信号进行特征提取的方法. 并通过三类地面活动数据采集进行对比识别实验, 分析了低采样率条件下地震信号特征提取方法的性能. 结果表明, 使用功率谱二次分析的特征提取方法能够降低网络通信能耗, 延长网络寿命, 提高系统目标识别的准确性.该方法已应用于秦始皇兵马俑野外文物安防系统, 经实践检验, 收到了良好的效果. 关键词： 地震信号 特征提取 功率谱二次处理 无线传感器网络     

一维空洞卷积神经网络的矿物光谱分类

矿物光谱综合反映了岩矿的物理化学特性、组分和内部结构特征,已被应用于岩矿识别研究。传统的矿物光谱分类方法需要先对矿物光谱进行预处理,再采用不同方法分析光谱特征,从而实现分类目的。但同时也会造成部分光谱信息丢失,导致最终分类精度不高且操作过程繁琐、效率低下,难以应对日益增长的大数据处理需求。因此,建立一个准确、高效的矿物光谱自动分类模型意义重大。卷积神经网络是应用最广泛的深度学习模型之一,它通过逐层抽取数据特征并组合形成高层语义信息,具有极强的模型表达能力,在光谱数据分析方面应用潜力巨大。针对矿物光谱数据的特点,提出了基于一维空洞卷积神经网络（1D-DCNN）的矿物光谱分类方法,利用空洞卷积神经网络提取光谱特征,采用反向传播算法结合随机梯度下降优化器调整模型参数,输出光谱分类结果,实现了矿物类别的端到端检测。该网络包含1个输入层、3个空洞卷积层、2个池化层、2个全连接层和1个输出层,采用交叉熵为损失函数,引入空洞卷积扩大滤波器感受野,有效避免光谱细节特征丢失。实验采集了白云母、白云石、方解石、高岭石四种矿物光谱,并通过添加噪声的方式进行数据增强,构建数量充足的矿物光谱样本用于神经网络模型训练与测试;探讨了卷积类型、迭代次数对模型分类结果的影响,并与多种传统矿物光谱分类方法进行对比,评价模型性能。实验结果表明,提出的1D-DCNN模型可实现矿物光谱快速准确分类,分类准确率达到99.32%,优于反向传播算法（BP）和支持向量机（SVM）,说明所提方法能够充分学习矿物光谱特征并有效分类,且模型具有良好的鲁棒性和可扩展性。该方法也可推广到煤炭、油气、月壤等其他领域光谱分类应用中。     

VCSEL-based gigabit IR-UWB link for converged communication and sensing applications in optical metro-access networks

We report on experimental demonstration of an impulse radio ultrawideband (IR-UWB) based converged communication and sensing system. A 1550-nm VCSEL-generated IR-UWB signal is used for 2-Gbps wireless data distribution over 800-m and 50-km single mode fiber links which present short-range in-building and long-reach access network applications. The IR-UWB signal is also used to simultaneously measure the rotational speed of a blade spinning between 18 and 30 Hz. To the best of our knowledge, this is the very first demonstration of a simultaneous gigabit UWB telecommunication and wireless UWB sensing application, paving the way forward for the development and deployment of converged UWB VCSEL-based technologies in access and in-building networks of the future.     

TERFDA: Tensor Embedding RF Domain Adaptation for varying noise interference

Radio frequency machine learning (RFML) can be loosely termed as a field that machine learning (ML) and deep learning (DL) techniques to applications related to wireless communications. However, traditional RFML basically assume that the data of training set and test set are independent and identically distributed and only a large number of labeled data can train a classification model which can effectively classify test set data. In other words, without enough training samples, it is impossible to learn an automatic modulation classifier that performs well in varying noise interference environment. Feature-based transfer learning minimizes the distribution difference between historical modulated signal data and new data by learning similarity-maximizing feature spaces. Therefore, in this paper, Dynamic Distribution Adaptation (DDA) is adopted to address the above challenges. We propose a Tensor Embedding RF Domain Adaptation (TERFDA) approach, which learns the latent subspace of the tensors formed by the time–frequency maps of the signals, so that use the multi-dimensional domain information of the signals to jointly learn the shared feature subspace of the source domain and the target domain, then perform DDA in the shared subspace. The experimental results show that under the modulated signal data, compared with the state-of-the-art DA algorithm, TERFDA has less requirements on the number of samples and categories, and has superior performance for confrontation the varying noise interference between source domain and target domain.     

神经网络的声场景自动分类方法*

声场景探察和自动分类能帮助人类制定应对特定环境的正确策略,具有重要的研究价值。随着卷积神经网络的发展,出现了许多基于卷积神经网络的声场景分类方法。其中时频卷积神经网络(TS-CNN)采用了时频注意力模块,是目前声场景分类效果最好的网络之一。为了在保持网络复杂度不变的前提下进一步提高网络的声场景分类性能,该文提出了一种基于协同学习的时频卷积神经网络模型(TSCNN-CL)。具体地说,该文首先建立了基于同构结构的辅助分支参与网络的训练。其次,提出了一种基于KL散度的协同损失函数,实现了分支与主干的知识协同,最后,在测试过程中,为了不增加推理计算量,该文提出的模型只使用主干网络预测结果。在ESC-10、ESC-50和UrbanSound8k数据集的综合实验表明,该模型分类效果要优于TS-CNN模型以及当前大部分的主流方法。     

基于小波系数图和卷积神经网络的太赫兹光谱物质识别

许多太赫兹光谱物质识别方法依靠寻找该物质在太赫兹波段范围内不同光谱表现出的不同特征来识别特定物质。吸收峰提取法是常用的光谱特征提取算法,但当光谱无明显特征吸收峰或峰位、峰值相近或难以识别时,难以利用吸收峰特征辨别物质。将机器学习和统计学习技术用于太赫兹光谱的识别中虽减少了吸收峰的干扰,但常常需要人为定义特征而导致分类误差。深度学习法能自动提取特征,但在识别前往往需要进行复杂的预处理操作,并且在特征提取的过程中容易丢失部分特征从而导致分类误差。针对以上问题,提出了一种基于小波系数图和卷积神经网络的太赫兹光谱识别方法。利用太赫兹光谱信号进行小波变换时,由于小波系数矩阵的每一行系数与原始光谱信号存在着对应关系,因此将太赫兹光谱的吸收系数通过小波变换在频率域上展开,能得到不同的二维的频率-尺度分布图,又称小波系数图。然后构造一个卷积神经网络(CNN)对小波系数图进行分类,可得到太赫兹光谱物质的分类结果。为了验证所提出算法的有效性,将三组小波系数图数据与原始光谱数据分别输入CNN、Support Vector Machin (SVM)、Multilayer Perceptron (MLP)三种不同的分类器作对比,从实验结果可以发现本文算法在三组数据中的识别率均达到了100%,说明相比于传统方法,本文方法能准确分类没有明显特征吸收峰的光谱,证明了使用卷积神经网络识别小波系数图的有效性。为了体现本文算法的优势,与小波脊线寻峰识别算法作对比,实验结果表明本文算法几乎不受峰频、峰位、峰值的影响,无论是识别不存在吸收峰的淀粉,还是识别相似度高的蔗糖和葡萄糖,都具有较高的识别率,分类准确率达97.62%,证明了所提算法的优越性。该算法为太赫兹光谱数据识别提供了一种新思路,同时也可以推广运用到其他谱图物质的识别中。     

Deep reinforcement learning based relaying for buffer-aided cooperative communications

The advances in deep reinforcement learning (DRL) have shown a great potential in solving physical layer-related communication problems. This paper investigates DRL for the relay selection in buffer-aided (BA) cooperative networks. The capability of DRL in handling highly-dimensional problems with large state and action spaces paves the way for exploring additional degrees-of-freedom by relaxing the restrictive assumptions around which conventional cooperative networks are usually designed. This direction is examined in our work by advising and analyzing advanced DRL-based BA relaying strategies that can cope with a variety of setups in multifaceted cooperative networks. In particular, we advise novel BA relaying strategies for both parallel-relaying and serial-relaying systems. For parallel-relaying systems, we investigate the added value of merging packets at the relays and of activating the inter-relay links. For serial-relaying (multi-hop) systems, we explore the improvements that can be reaped by merging packets and by allowing for the simultaneous activation of sufficiently-spaced hops. Simulation results demonstrate the capability of DRL-based BA relaying in achieving substantial improvements in the network throughput while the adequate design of the reward/punishment in the learning process ensures fast convergence speeds.     

Analysis of mobility impact on interference in cognitive radio networks

Cognitive radio (CR) technology seems to be a promising candidate for solving the radio frequency (RF) spectrum occupancy problem. CRs strive to utilize the white holes in the RF spectrum in an opportunistic manner. Because interference is an inherent and a very critical design parameter for all sorts of wireless communication systems, many of the recently emerging wireless technologies prefer smaller size coverage with reduced transmit power in order to decrease interference. Prominent examples of short-range communication systems trying to achieve low interference power levels are CR relays in CR networks and femtocells in next generation wireless networks (NGWNs). It is clear that a comprehensive interference model including mobility is essential especially in elaborating the performance of such short-range communication scenarios. Therefore, in this study, a physical layer interference model in a mobile radio communication environment is investigated by taking into account all of the basic propagation mechanisms such as large- and small-scale fading under a generic single primary user (PU) and single secondary user (SU) scenario. Both one-dimensional (1D) and two-dimensional (2D) random walk models are incorporated into the physical layer signal model. The analysis and corresponding numerical results are given along with the relevant discussions.