Search

While deep learning (DL) has been widely employed for the classification of hyperspectral imagery (HSI), many scenarios arise in practice in which too few labeled samples exist to effectively train the networks. Few-shot learning has been recently used to deploy classifiers trained on source-domain datasets comprising a large number of labeled samples to datasets from a target domain with only few labeled samples. However, most techniques in this vein effectively assume that the source and target domains possess the same data distribution, whereas the distributions between the two domains often differ widely in practice. Adversarial domain adaption driven by prototype classifiers deployed independently in the source and target domains is proposed to handle such differing source and target distributions, while an attention-based feature extractor with residual skip connections is developed in order to weight spectral bands according to their importance to the hyperspectral classification task. Experimental results demonstrate improved performance for the proposed few-shot-learning framework relative to both fully supervised classifiers as well as other few-shot techniques. [ABSTRACT FROM AUTHOR]

Convolutional networks have been widely used for the classification of hyperspectral images; however, such networks are notorious for their large number of trainable parameters and high computational complexity. Additionally, traditional convolution-based methods are typically implemented as a simple cascade of a number of convolutions using a single-scale convolution kernel. In contrast, a lightweight multiscale convolutional network is proposed, capitalizing on feature extraction at multiple scales in parallel branches followed by feature fusion. In this approach, 2-D depthwise convolution is used instead of conventional convolution to reduce network complexity without sacrificing classification accuracy. Furthermore, multiscale channel attention (MSCA) is also employed to selectively exploit discriminative capability across various channels. To do so, multiple 1-D convolutions with varying kernel sizes provide channel attention at multiple scales, again with the goal of minimizing network complexity. Experimental results reveal that the proposed network not only outperforms other competing lightweight classifiers in terms of classification accuracy, but also exhibits a lower number of parameters as well as significantly less computational cost. [ABSTRACT FROM AUTHOR]

Deep learning is being increasingly employed for hyperspectral classification, although such use is often predicated on the availability of a sufficiently large set of labeled samples for training. To improve classification performance under a limited training-set size, a semi-supervised network with end-to-end local–global active learning (AL) based on graph convolutional networks (GCNs) is proposed. The proposed AL extracts both global as well as local graph-based features to gauge the discriminative information in unlabeled samples, while semi-supervised classification expands the training set of a fully supervised classifier by attaching pseudo-labels to high-confidence unlabeled samples. Experimental results demonstrate that the proposed network outperforms not only other approaches to semi-supervised classification but also several existing fully supervised methods. The source code of this method can be found at https://github.com/XtaoS/semi-LG-AGCN. [ABSTRACT FROM AUTHOR]

Light detection and ranging (LiDAR) has been widely used in autonomous vehicles for perception and localization. However, the cost of a high-resolution LiDAR is still prohibitively expensive, while its low-resolution counterpart is much more affordable. Therefore, using low-resolution LiDAR for autonomous driving is an economically viable solution, but the point cloud sparsity makes it extremely challenging. In this letter, we propose a two-stage neural network framework that enables 3-D object detection using a low-resolution LiDAR. Taking input from a low-resolution LiDAR point cloud and a monocular camera image, a depth completion network is employed to produce dense point cloud that is subsequently processed by a voxel-based network for 3-D object detection. Evaluated with KITTI dataset for 3-D object detection in bird-eye view (BEV), the experimental result shows that the proposed approach performs significantly better than directly applying the 16-line LiDAR point cloud for object detection. For both easy and moderate cases, our 3-D vehicle detection results are close to those using 64-line high-resolution LiDARs. [ABSTRACT FROM AUTHOR]

In this article, a 3-D geometry-based stochastic model (GBSM) is proposed for millimeter wave (mmWave) massive multiple-input multiple-output (MIMO) unmanned aerial vehicle (UAV) channels. The proposed model is the first mmWave massive MIMO UAV two-cylinder GBSM that enables to jointly model the channel space-time-frequency non-stationarity by a novel variable parameter-based method. In this novel method, key parameters of UAV channels are assumed to vary in space, time, and frequency domains, and the effect of the unique UAV-related parameters is further taken into account, such as the UAV’s altitude, velocity, and moving directions. Based on the proposed model, some statistical properties are derived, including the time-variant transfer function (TVTF), space-time-frequency correlation function (STF-CF), Doppler power spectral density (PSD), and the standard deviation of Doppler frequency on antenna arrays. Simulation results show that the channel non-stationarity in space, time, and frequency domains can be captured, and the aforementioned UAV channel-related parameters have a great impact on channel statistics. Finally, the utility of the proposed model is validated by the excellent agreement between simulation results and measurements. [ABSTRACT FROM AUTHOR]

Unmanned aerial vehicle (UAV)-enabled backscatter communications (BackComm) is deemed to be a vital technique enabling the data transmission over massive battery-less devices for Internet of Things (IoT). However, the UAV-enabled Backcomm suffers from information leakage due to the broadcasting nature of wireless channels. To cope with the security issue, in this paper, a UAV-enabled multi-user secure BackComm system is developed using analog beamforming (ABF) and randomized continuous wave (RCW) techniques, where the multiple users are supported by the multi-carrier RCW over a single low-complexity radio frequency (RF) chain. By exploiting the RCW transmitted towards backscatter, the eavesdropping link can be eroded without any specific jamming signals. The closed-form of the secrecy rate is studied with the approximations, which is then maximized by jointly optimizing the beamforming together with the UAV’s location and the RCW settings. Simulation results are carried out to confirm the accuracy of the proposed approximation, while the convergence behavior of the optimization algorithm is analyzed. As a result, it can be shown that the secrecy rate can be significantly improved compared with the benchmark schemes. [ABSTRACT FROM AUTHOR]

We study the problem of simultaneous search for multiple targets over a multidimensional unit cube and derive fundamental resolution limits of non-adaptive querying procedures using the 20 questions estimation framework. The performance criterion that we consider is the achievable resolution, which is defined as the maximal $L_\infty $ norm between the location vector and its estimated version where the maximization is over all target location vectors. The fundamental resolution limit is defined as the minimal achievable resolution of any non-adaptive query procedure, where each query has binary yes/no answers. We drive non-asymptotic and second-order asymptotic bounds on the minimal achievable resolution, using tools from finite blocklength information theory. Specifically, in the achievability part, we relate the 20 questions problem to data transmission over a multiple access channel, use the information spectrum method by Han and borrow results from finite blocklength analysis for random access channel coding. In the converse part, we relate the 20 questions problem to data transmission over a point-to-point channel and adapt finite blocklength converse results for channel coding. Our results extend the purely first-order asymptotic analyses of Kaspi et al. (ISIT 2015) for the one-dimensional case: we consider channels beyond the binary symmetric channel and derive non-asymptotic and second-order asymptotic bounds on the performance of optimal non-adaptive query procedures. [ABSTRACT FROM AUTHOR]

Automatically generating an accurate and meaningful description of an image is very challenging. However, the recent scheme of generating an image caption by maximizing the likelihood of target sentences lacks the capacity of recognizing the human–object interaction (HOI) and semantic relationship between HOIs and scenes, which are the essential parts of an image caption. This article proposes a novel two-phase framework to generate an image caption by addressing the above challenges: 1) a hybrid deep learning and 2) an image description generation. In the hybrid deep-learning phase, a novel factored three-way interaction machine was proposed to learn the relational features of the human–object pairs hierarchically. In this way, the image recognition problem is transformed into a latent structured labeling task. In the image description generation phase, a lexicalized probabilistic context-free tree growing scheme is innovatively integrated with a description generator to transform the descriptions generation task into a syntactic-tree generation process. Extensively comparing state-of-the-art image captioning methods on benchmark datasets, we demonstrated that our proposed framework outperformed the existing captioning methods in different ways, such as significantly improving the performance of the HOI and relationships between HOIs and scenes (RHIS) predictions, and quality of generated image captions in a semantically and structurally coherent manner. [ABSTRACT FROM AUTHOR]

To extend the conventional wireless sensor networks (WSNs) to support wider applications such as intruder detection and border security monitoring, active radar sensors are introduced into WSNs to further enhance their capability, thus forming wireless radar sensor networks (WRSNs). To improve the network efficiency, the technology of integrated sensing and communication (ISAC) can be applied to co-design the sensing and communication functionalities of radar sensors. Since the cooperative operations of WRSNs require effective information interaction among radar sensors, data dissemination techniques need to be investigated, which become even more critical in desolate areas without the coverage of the base stations (BSs). Therefore, in this paper, a duty cycling mechanism is applied to the network to enhance the usage of WRSNs and support data dissemination, where a storage node is deployed to store the data spreading from radar sensors and a mobile data collector is employed to collect the data from the storage node periodically. Then, the epidemic theory, as an innovative tool for modeling data dissemination, is adopted to analyze the performance of WRSNs. After epidemiological modeling, the density of radar sensors is optimized by the epidemiological analytical method to maximize the throughput of the storage node by jointly considering the functions of radar detection and communication. Simulation results validate the accuracy of analysis, which also show the efficiency of the optimization for data dissemination. [ABSTRACT FROM AUTHOR]

In remote sensing scene classification (RSSC), features can be extracted with different spatial frequencies where high-frequency features usually represent detailed information and low-frequency features usually represent global structures. However, it is challenging to extract meaningful semantic information for RSSC tasks by just utilizing high- or low-frequency features. The spatial composition of remote sensing images (RSIs) is more complex than that of natural images, and the scales of objects vary significantly. In this article, a multiscale feature fusion covariance network (MF2CNet) with octave convolution (Oct Conv) is proposed, which can extract multifrequency and multiscale features from RSIs. First, the multifrequency feature extraction (MFE) module is used to obtain fine-grained frequency features by Oct Conv. Then, the features of different layers in MF2CNet are fused by the multiscale feature fusion (MF2) module. Finally, instead of using global average pooling (GAP), global covariance pooling (GCP) extracts high-order information from RSIs to capture richer statistics of deep features. In the proposed MF2CNet, the obtained multifrequency and multiscale features can effectively improve the performance of CNNs. Experimental results on four public RSI datasets show that MF2CNet has advantages in RSSC over current state-of-the-art methods. The source codes of this method can be found at https://github.com/liuqingxin-chd/MF2CNet. [ABSTRACT FROM AUTHOR]

The next generation multiple access (NGMA) schemes are considered to support massive access for a large number of devices, which motivates us to develop a low-complexity approach for next generation systems. Since the generalized spatial modulation (SM) can be adopted to the system, a number of compressive sensing (CS) reconstruction algorithms are deployed for the detection of sparse signals, while the complexity of CS-based approaches is proportional to the number of antennas. In order to decrease the complexity, we propose a two-stage approach to detect sparse signals, where the received signals are divided into groups. Then, the activity variables of aggregated signals are decided and the sparse signal detection is carried out at the signals belonging to active groups. During the activity variable detection, the variational inference algorithm is applied to determine the activity variables. Moreover, in order to analyze the performance of activity variable detection, the $J$ -divergence is proposed to measure the distance between the distributions, while the approximate expression of $J$ -divergence is derived. Simulation results show that the proposed approach is able to provide good detection performance with low complexity. In addition, the $J$ -divergence is confirmed to be useful as an evaluation metric to measure the detection performance. [ABSTRACT FROM AUTHOR]

The intelligent transportation has been extensively investigated as an enabling technology for ubiquitous data processing and content sharing among vehicles and terrestrial infrastructures. In intelligent transportation systems, numerous vehicles and infrastructures are connected for information and data sharing to enable different operations. Since there are some urban areas that face the traffic congestion or cannot be well served, space-air-ground integrated networks (SAGIN) can be carried out to provide continuous network connectivity for vehicles. In particular, unmanned aerial vehicles (UAVs) are deployed as data collectors to receive data packets from vehicles due to the advantages of high mobility and low operating cost. It is noteworthy that the information freshness is critical to enable services for timely decision, e.g., autonomous driving and accident prevention. In this paper, we develop UAV-aided intelligent transportation systems to enhance the usage of vehicular networks and support low latency vehicular services, where the concept of age-of-information (AoI) is adopted to measure the freshness of data packets of vehicles. Then, the performance of UAV-aided intelligent transportation systems is analyzed in terms of the average AoI. In addition, the deployment of multiple UAVs is optimized to minimize the average peak AoI according to the traffic intensity of vehicles under seamless coverage, finite queue, and coverage probability constraints. To this end, the deployment optimization problem is formulated as a multi-constrained non-convex optimization problem and solved by considering each soft constraint separately. Simulation results show that our proposed system can provide timely data transmission. [ABSTRACT FROM AUTHOR]

Visible light positioning and computer vision-based localization have the potential to be cost-effective technologies for accurate indoor localization. However, the feasibility of existing methods in this domain is limited. In this paper, a novel visible light communication (VLC)-assisted perspective-four-line algorithm (V-P4L) is proposed for practical indoor localization. The basic idea of V-P4L is to jointly use VLC and computer vision techniques to achieve high localization accuracy regardless of LED height differences. In particular, the space-domain information is first exploited to estimate the orientation and coordinate information of a single rectangular LED luminaire in the camera coordinate system based on plane geometry theory and solid geometry theory. Then, by using time-domain information transmitted by VLC and the estimated luminaire information, the proposed V-P4L can estimate the position and pose of the camera using single-view geometry theory and the linear least square (LLS) method. To further mitigate the effect of height differences among LEDs on localization accuracy, a correction algorithm based on the LLS method and a simple optimization method is proposed. Due to the combination of time- and space-domain information, V-P4L can achieve accurate localization using a single luminaire without limitation on the correspondences between the features and their projections in conventional perspective-n-line (PnL) algorithms. Simulation results show that the position error caused by the proposed V-P4L algorithm is always less than 15 cm and the orientation error is always less than 4° using popular indoor luminaires. Experimental results with real hardware show that the average position error is less than 3 cm under both similar and different heights for the LEDs. [ABSTRACT FROM AUTHOR]

The Internet of Things (IoT) has been extensively investigated as an enabling technology for situation awareness and data collection. Since power-limited or battery-free IoT devices may have limited transmission coverage, unmanned aerial vehicles (UAVs) can be employed as mobile edge computing (MEC) servers to collect and process data packets from multiple IoT devices due to their high mobility and low operating cost. However, data transmission becomes vulnerable when eavesdroppers exist, as those IoT devices may have limited computing power for encryption. Moreover, it is noteworthy that the identity and location information of IoT devices themselves are also crucial to particular implementations (e.g., regional situation awareness), especially when massive IoT devices are involved using random access schemes. In this article, a secure structure is proposed as a promising strategy to support UAV-aided IoT networks, where the identity and location information of multiple IoT devices are utilized as the key to encrypt data packets collected from IoT networks. In the new structure, the trajectory planning of UAVs is developed to minimize the energy consumption among multiple clusters of a UAV, while the deployment optimization is used to maximize the secrecy capacity in each cluster, and the beam hopping is carried out to encrypt multi-access IoT devices within a coverage of a UAV. After introducing the structure, some key technologies are presented, and then future works for the secure UAV-aided IoT network are discussed. [ABSTRACT FROM AUTHOR]

A novel single-channel direction finding (DF) method with spatial spectrum estimation is proposed based on 1 bit time-modulated array (TMA). Its performance is analyzed from theoretical derivation to numerical simulations and compared with other existing DF methods. The proposed DF method is implemented as follows. First, the received sideband signals from the 1 bit modulated single-channel signal are deduced, followed by the detailed analysis of the received noise distribution. Second, the harmonic recovery (HR) method with the $l_{2} $ -norm approximation is utilized to recover the original array signals. Finally, conventional spatial spectrum estimation methods are easily applied for high-accuracy DF. Meanwhile, The DF accuracy of the proposed 1 bit method is analyzed under the Gaussian noise model, and its Cramér–Rao lower bounds (CRLBs) are deduced for the first time and compared with conventional arrays and other DF TMA systems. Numerical simulations are provided to show that the proposed 1 bit TMA system and HR method can achieve better accuracy performance over the existing TMA systems when the circuit noise is predominant in the received noises. An $S$ -band 1 bit TMA system is fabricated and implemented to experimentally verify the feasibility of the proposed method under circumstances of a single incident source and two coherent sources. [ABSTRACT FROM AUTHOR]

In 5G massive machine-type communication (mMTC), power-limited or battery-free parasite devices such as radio frequency identification (RFID) tags, can use the transmitted signals from host devices as ambient signals for backscatter communications to send information to a base station (BS). Unmanned aerial vehicles (UAVs) can be employed as host devices to help transmissions of parasite devices due to the advantages of high mobility and low operating cost. In this paper, we propose a signal detection approach based on the central limit theorem to detect the presence of parasite devices and separate parasite signals from host signals. Then, closed-form expressions for the probability of error detection and the bit error rate (BER) are derived. Moreover, the trajectory planning of multiple UAVs is optimized with the consideration of minimizing the energy consumption of UAV swarms to serve parasite devices. Theoretical and simulation results show that our proposed method provides good detection performance for parasite devices. It also shows that the trajectory planning of multiple UAVs is optimized. [ABSTRACT FROM AUTHOR]

A high-coverage algorithm termed enhanced camera assisted received signal strength ratio (eCA-RSSR) positioning algorithm is proposed for visible light positioning (VLP) systems. The basic idea of eCA-RSSR is to utilize visual information captured by the camera to estimate first the incidence angles of visible lights. Based on the incidence angles, eCA-RSSR utilizes the received signal strength ratio (RSSR) calculated by the photodiode (PD) to estimate the ratios of the distances between the LEDs and the receiver. Based on an Euclidean plane geometry theorem, eCA-RSSR transforms the ratios of the distances into the absolute values. In this way, eCA-RSSR only requires three LEDs for both orientation-free 2D and 3D positioning, implying that eCA-RSSR can achieve high coverage. Based on the absolute values of the distances, the linear least square method is employed to estimate the position of the receiver. Therefore, for the receiver having a small distance between the PD and the camera, the accuracy of eCA-RSSR does not depend on the starting values of the non-linear least square method and the complexity of eCA-RSSR is low. Furthermore, since the distance between the PD and camera can significantly affect the performance of eCA-RSSR, we further propose a compensation algorithm for eCA-RSSR based on the single-view geometry. Experiment results show that positioning errors of less than five centimeters is achievable for eCA-RSSR. Simulation results show that eCA-RSSR can achieve 80th percentile accuracy of about four centimeters and can improve the coverage ratio at low cost. [ABSTRACT FROM AUTHOR]

In recent years, the Internet of Things (IoT) has been extensively investigated as an enabling technology to support massive devices or sensors. With increasing demands of seamless connectivity of IoT, space information networks (SINs) are adopted to provide reliable communications for IoT devices at any time and anywhere. However, the conventional SIN suffers from different malicious attacks (e.g., jamming attack, eavesdropping attack) due to its highly exposed satellite-ground links with fixed spot beams. Those attacks may cause severe security problems to SINs. To overcome these drawbacks, in this article, a secure architecture is proposed as a promising strategy to improve the security of SINs, where relays are deployed together with hopped beams to relieve the jamming attack of the uplink and combat the eavesdropping attack of the downlink. In the proposed new structure, by utilizing spatial diversity, inter-satellite beam hopping is developed to build jamming-free uplink channels, while relay selection is carried out to form eavesdropping-free downlink channels. Meanwhile, relay deployment optimization is studied to maximize the system efficiency, and data aggregation is adopted to encrypt the data packets of IoT devices at relays. As a result, the security of SINs can be strongly improved in the existence of jamming and eavesdropping threats. Furthermore, future works for the secure SIN are also discussed. [ABSTRACT FROM AUTHOR]

The time-modulated array (TMA)-based space-division multiple access (SDMA) approach is studied and verified, including the performance analysis and experimental verification. It utilizes the multibeam characteristic of TMA to implement the SDMA. For the uplink, the main power of signals with different incident angles would be redistributed to different harmonic components after time modulation. In this case, it plays a key role in this system to determine the harmonic order, which is selected to detect the received signal. A novel harmonic selection method for the uplink TMA-based SDMA systems is proposed by maximizing the received signal power. Furthermore, a closed-form expression of the spectral efficiency is deduced to verify the advantage of the simple harmonic selection scheme. Numerical results show that the TMA-based SDMA system can significantly improve the spectral efficiency compared with traditional single radio frequency (RF) chain systems. Meanwhile, it does not increase the number of RF chains. An L-band eight-element TMA is fabricated and tested to validate the feasibility of the TMA-based SDMA method. [ABSTRACT FROM AUTHOR]

In 5G and beyond wireless scenarios, it is expected to support tremendous amount of communicating machines. With an exponential increase of machine-to-machine (M2M) system deployments, efficient approaches to massive access by a large number of devices are to be studied. In this paper, we propose a massive multiple-input multiple-output (MIMO) based grant-free random access (RA) with resolution of preamble collision for massive access. Based on the channel hardening and favorable propagation characteristics of massive MIMO, collided signals processed at the base station (BS) can be viewed as a variation of superposition modulation, and are to be recovered by successive interference cancellation (SIC) techniques. Besides, taking into consideration the effect of pass loss and fractional power control (FPC), analytic expressions of success probability of the proposed collision resolution with conjugate beamforming (CB) and zero-forcing beamforming (ZFB) are derived. With simulation results, we verify the analyses and show that the proposed protocol can resolve most preamble collisions. [ABSTRACT FROM AUTHOR]