Search

In this paper, we investigate the problem of sum rate maximization by guaranteeing the quality of service (QoS) of both uplink and downlink in cell-free with network-assisted full Duplexing (NAFD), where the users operate in half-duplex mode and access points (APs) operate in either full-duplex mode or half-duplex mode. In the considered network, the central processor unit (CPU) sends the compressed beamformed signals to the transmit-APs (T-APs) over the downlink fronthaul, and the T-APs forward the signals to downlink users. At the same time, the receive-APs (R-APs) compress the signals transmitted by uplink users and forward them to the CPU via uplink fronthaul. We aim to maximize the spectral efficiency and the number of users that should be admitted by the network, where users’ requirements of both downlink and uplink signal-to-interference-plus-noise (SINR) constraints, fronthaul capacity constraints, energy harvesting constraints and simultaneous wireless information and power transfer (SWIPT) ratio design are considered. A successive convex approximation-based algorithm is proposed to solve the highly coupled problem, which is guaranteed to converge to the Karush-Kuhn-Tucker (KKT) conditions. We conduct a comprehensive comparison between the NAFD scheme and the traditional co-frequency co-time full duplex (CCFD) scheme and time division duplex (TDD) scheme and offer some valuable opinions about the system design.

Underwater environment is complex and random. The images obtained from underwater by sonar always have uneven background gray distribution and fuzzy details of boundary. Hence the low-quality sonar images need to be enhanced before analysis. This paper presents a sonar image enhancement algorithm based on the approximation spaces of random sets. First of all, the knowledge representation of the underwater image is constructed by the approximation spaces of random sets. According to the background knowledge, the image is divided by the upper and lower approximation space of the random set. Then the optimal partition is obtained according to the approximate equivalence relation of the upper and lower approximation. Based on the optimal partition, an improved dark channel theory is presented to enhance each region of the image. After that, sonar images with different backgrounds are used to test the proposed method. The experimental results show that the gray distribution of the sonar image enhanced by this algorithm is more uniform and the boundary details are clearer. The proposed algorithm has the advantage of solving the optimal division for the set of pixels with approximate grayscale. Moreover, the proposed algorithm can get better image enhancement effect in the premise of maintaining the texture of the sonar images.

Arbitrarily-oriented object detection is a challenging task. Since the object orientation in remote sensing images is arbitrary, using horizontal bounding boxes will lead to low detection accuracy. Existing regression-based rotation detectors can lead to the problem of boundary discontinuity. In this paper, we propose a remote sensing image object detection method based on angle classification that uses rotation detection bounding boxes with angle information to detect objects. Specifically, we incorporate the neural architecture search framework with feature pyramid network (NAS-FPN) module in a dense detector (RetinaNet) and use a binary encoding method in angle classification. This method reduces the background influence, so that there is almost no overlap between detection boxes. Based on the angles of the detection boxes, we can infer the information of the motion direction of the target and further determine the motion trajectory of the target. We conducted ablation experiments on a large publicly available for object detection in an aerial imagery (DOTA) to verify the effectiveness of each module in the method and compared the method with several other detection methods. The experimental results demonstrate the effectiveness of our method.

Network-assisted full duplex (NAFD) is a unified duplex technique for large-scale distributed antenna systems (DAS) with remote antenna units operating in either half-duplex mode or full-duplex mode and all connecting to central processing unit via backhaul links. Downlink sparse beamforming and uplink power control are effective methods to reduce cross-link interference involved in NAFD with finite-capacity backhaul. We study joint downlink beamforming and power control for NAFD by maximizing the aggregated uplink and downlink spectral efficiency subject to quality-of-service constraints and backhaul constraints. Two approaches have been proposed to solve the optimization problem, where the first approach converts the objective function to the difference of two convex functions with semi-definite relaxation, and then, an iterative SDR-block coordinate descent method is applied to solve the problem. The second method is based on sequential parametric convex approximation. Simulation results have shown that the proposed algorithms yield a higher spectral efficiency gain compared to the traditional time-division duplex scheme.

To achieve the expected attractive potential of massive multiple-input multiple-output (MIMO), accurate match between the receiver and the actual channel is necessary. Due to the channel aging, this kind of mismatch appears in the vast majority of practical propagations. Consequently, the impact of channel aging for massive MIMO is investigated in this study. The authors first propose an equivalent estimated channel model by considering the inevitable time-varying nature of channels in realistic mobile communications. Then, they obtain the achievable uplink rate based on the proposed equivalent channel models. In particular, the closed-form expression of the asymptotic uplink rate is derived. Different from the existing works, this study considers the joint multi-cell minimum-mean-squared-error (MMSE) channel estimator and MMSE receiver for massive MIMO systems over aging channels. Through the simulation, it demonstrates that adopting massive MIMO can compensate for the performance loss caused by channel aging. In addition, this pstudy gives some insights into the system design: to achieve the best performance, system with higher-mobility terminals should be operated with more pilot resources, and it can support more users, moreover, the optimal number of users supported in a cell is less than half of the coherence time interval.

This study focuses on the spectral efficiency of massive multiple-input–multiple-output (MIMO) systems over Ricean fading channels. Assuming the channel reciprocity of uplink and downlink transmissions, the channels are estimated through uplink pilot sequence with least minimum mean square error (MMSE) estimator, which facilitates the investigation of equivalent channel model. With the equivalent channel model, adopting joint MMSE receiver, the uplilnk achievable sum-rate is studied, and finally, the asymptotic expression of spectral efficiency is obtained. Based on the proposed asymptotic spectral efficiency, numerical results give some guidelines about system parameters design: the authors should choose a proper length of the block fading channels to reduce the effect of pilot overhead, and an optimal length of pilot sequence always exists to guarantee a satisfying system performance in terms of spectral efficiency. Furthermore, the simulation results show that when fixing the number of users, shrinking the cell serving area and increasing the number of antennas can improve the system performance.

Crack detection is important for evaluating pavement conditions. In this paper, we propose an automatic pavement crack detection method called as U2CrackNet. The architecture of U2CrackNet is a two-level nested U-structure which is an encoding and decoding architecture. Firstly, the crack features are extracted by the encoding layer so that the preliminary effective feature layer is obtained. Subsequently, the encoder and decoder are connected by an atrous spatial pyramid pooling (ASPP) model. The atrous convolution with different expansion rates can be used to capture the multi-scale crack information. Besides, in order to make the network pay more attention to the features of cracks, the crack feature map channels are given different weights through the effective channel attention mechanism after upsampling in the decoding layer. Finally, the crack saliency probability maps generated by each layer of the network are fused into the final crack saliency map through cascade operation. The proposed method has been evaluated on an expanded pavement crack dataset containing 8700 images. The experimental results demonstrate that the proposed U2CrackNet can obtain more clear and continuous cracks. Specifically, the precision, accuracy, F1-score, and Mean Intersection over Union (MIoU) are 89.51%, 98.95%, 81.45%, and 69.19%, respectively.

Underwater organisms are an important part of the underwater ecological environment. More and more attention has been paid to the perception of underwater ecological environment by intelligent means, such as machine vision. However, many objective reasons affect the accuracy of underwater biological detection, such as the low-quality image, different sizes or shapes, and overlapping or occlusion of underwater organisms. Therefore, this paper proposes an underwater biological detection algorithm based on improved Faster-RCNN. Firstly, the ResNet is used as the backbone feature extraction network of Faster-RCNN. Then, BiFPN (Bidirectional Feature Pyramid Network) is used to build a ResNet–BiFPN structure which can improve the capability of feature extraction and multi-scale feature fusion. Additionally, EIoU (Effective IoU) is used to replace IoU to reduce the proportion of redundant bounding boxes in the training data. Moreover, K-means++ clustering is used to generate more suitable anchor boxes to improve detection accuracy. Finally, the experimental results show that the detection accuracy of underwater biological detection algorithm based on improved Faster-RCNN on URPC2018 dataset is improved to 88.94%, which is 8.26% higher than Faster-RCNN. The results fully prove the effectiveness of the proposed algorithm.

In this paper, we investigate duplex mode selection and transceiver design for cell-free massive multiple-input multiple-output (MIMO) with network-assisted full duplexing (NAFD), where the remote antenna units (RAUs) simultaneously serve both uplink and downlink users on the same time-frequency resource, and the joint transmission and reception are done at the central processing unit. We consider that each antenna can operate in three modes, i.e., uplink reception, downlink transmission and sleep. In particular, we model the problem as a mixed-integer optimization problem to maximize the aggregated spectral efficiency (SE) of downlinks and uplinks, where the quality-of-service constraints and power budget constraints are considered. Since the uplinks and downlinks are highly coupled, to address the design problem, we resolve the combinatorial problem by a series of nonconvex-convex approximate methods, such as equivalent formulations, iterative success convex approximations, and binary relaxations. In particular, a two-stage strategy is proposed to solve the optimization problem. In the first stage, we fix the mode selection vectors and then obtain the optimal transceivers with such a fairness strategy. In the second stage, with the transceiver parameters obtained from the first stage, we further optimize the duplex mode of each antenna. The algorithms of the two stages will alternately run in several loops until convergence. Numerical results indicate that the proposed solution can obtain SE performance that is close to the optimal exhaustive search solution and yields a higher SE gain compared with the traditional fixed-mode duplex scheme.

Considering aging channel effect, the joint processing of pilot and data symbols scheme is investigated. Through the analysis of mutual information, the authors decompose the lower bound of the achievable rate into three terms with explicit physical meanings, and obtain the closed-form expression of the achievable rate. In the derivation of the closed-form expressions, this Letter proposes two novel joint processing algorithms. Through simulation, the joint processing is found to provide a better performance due to the extra information captured by received data signals. In addition, the joint processing scheme is shown to overcome the saturation area issue which further facilitates the application of joint processing. Moreover, the proposed algorithms are shown to enhance the performance gain comparing with the traditional one-step prediction method. The channel aging effect on joint processing is further discussed, remarkably, when the terminal moves fast, the joint processing can still support a relatively satisfying performance.

Recently, a novel network architecture for a distributed antenna system with interlaced clustering has been proposed to mitigate the cell-edge problem. Under this network architecture, we propose a more practical implementation for ultradensely deployed remote antenna units (RAUs) with large numbers of users. Furthermore, we focus on the user selection (USC) and sparse beamforming technologies to optimize the weighted sum rate (WSR) with both backhaul and power constraints. First, we divide each cluster pattern (CP) into several adaptive cells, where RAUs are connected to a central processor via finite-capacity backhaul links. Aiming at reducing the computational complexity for large numbers of users and RAUs, we solve the original problem with two steps. In the first stage, we propose an efficient USC algorithm to find the largest user subset that satisfies the quality of service requirement. In the second stage, we adopt a CP-based weighted sum minimum mean square error algorithm to optimize the WSR problem for the selected users in the previous stage. Moreover, two decomposition algorithms, named primal decomposition and dual decomposition, are exploited to further reduce the computational complexity. Furthermore, based on the adaptive cells, we provide a low-complexity alternating optimization method for sparse beamforming. Finally, simulation results show that the proposed algorithms can achieve a significant performance gain on edge-user rates without losing much performance gain. At the same time, the backhaul information exchange is largely reduced, and approximately 90% of the RAUs consumes less than 66.7% of backhaul for each RAU.

Besides the pilot symbols, the received data signals also contain some extra channel state information (CSI), which can be exploited to further improve the channel estimation accuracy. Considering a multi-cell multi-user massive multiple input multiple output (MIMO) with Ricean fading, the spectral efficiency for joint processing of the pilot and data strategy is investigated. Through the mutual information analysis of the multi-cell multi-user massive MIMO system, the intractable expression for the ergodic spectral efficiency is lower bounded and decomposed into three terms, where the explicit physical meanings of them are discussed in detail. Facilitated by the decomposing operation, an asymptotic closed-form expression of the spectral efficiency is obtained, and it is proven to be accurate. Comparing with separate processing, numerical results show that the joint processing outperforms the separate processing in spectral efficiency since the additional CSI is captured. The performance advantage for joint processing over separate processing becomes obvious as the coherence block duration of channels (T) increases. For a long enough T, joint processing eventually approaches the case with perfect CSI in terms of spectral efficiency. In addition, the inter-cell interference heavily reduces the system performance of joint processing when T is relatively small.

To cope with the growing trend of asymmetric data traffic, the bidirectional dynamic networks (BDNs) dynamically allocate the number of uplink and downlink remote radio heads (RRHs), which facilitates simultaneous uplink and downlink communications. In this study, the authors derive the asymptotic approximations of the achievable uplink and downlink rates using maximum ratio transmission precoder and maximum ratio combination receiver, as the RRH antenna number ( M ) approaches infinity. Considering an optical fibre connected backhaul network, a practical power consumption model is presented to study the system energy efficiency (EE). Based on the asymptotic analysis, they exploit the power scaling laws that both the uplink and downlink powers should scale down to 1/ M to maintain a desirable uplink or downlink rate. Numerical results verify that when M is large, the BDN system outperforms the dynamic time division duplex system in both the spectral efficiency and EE.

To cope with the ongoing trend in data traffic asymmetry of uplink (UL) and downlink (DL) transmissions, bidirectional dynamic networks (BDNs) have been proposed to facilitate simultaneous UL and DL communications. Moreover, the large-scale distributed antenna system (L-DAS) is considered to improve the spectral efficiency (SE). Aiming at maximizing the SE in the BDN with the L-DAS, in this paper, we propose a novel distributed antenna (DA) clustering strategy named flexible antenna clustering (FAC) to allow each user to choose the most effective DAs. This also provides a low-complexity solution to solve the antenna clustering problem in the L-DAS. In FAC, the operation mode (UL or DL transmission) of the DAs can be flexibly changed, which is determined by the baseband processor units. By taking both the homogeneous and heterogeneous interferences into consideration, we propose two novel metrics for the user and DA selection orders. To the best of our knowledge, this is the first time that the user selection order is considered in solving the antenna clustering problem. Compared with the time division duplex system, our FAC strategy used in the BDN is verified to have improved efficiency in achieving better SE performance. Therefore, the BDN with the L-DAS is suitable in practical communication systems thanks to its SE gain. Based on these results, we further provide some specific suggestions for the practical network configuration.

In this letter, a transmit power control architecture based on the recurrent neural network (RNN) is proposed to solve the power allocation problem for the interference channel to maximize the weighted sum-rate. Compared with the conventional power control schemes, such as weighted minimization mean squared error (WMMSE), which require a considerable number of computations, the RNN-based scheme not only requires much lower computational complexity, but also can well learn and characterize inter-user relationship in the interference channel. The simulation results show that the proposed scheme achieves the sum-rate close to the WMMSE-based scheme with much lower complexity and obtains higher sum-rate than the deep neural network (DNN)-based scheme with much less model parameters.

The Optical Character Recognition (OCR) technology is widely used in intelligent identification of bank cards, since it can improve the work efficiency and user experience in mobile payment. Conventional methods have the problems of low recognition rate and location accuracy. Therefore, an automatic location and recognition method for bank card number is proposed. Firstly, novel Connected Text Proposal Network (CTPN) algorithm is improved to locate the bank card number. Then, the Convolutional Recurrent Neural Networks (CRNN) algorithm is optimized to identify the card number. Some experimental results show that the method has a high positioning accuracy and recognition rate.

We consider the uplink transmission in vehicle-to-infrastructure (V2I) system, where the roadside units (RSUs) are equipped with a large number of antennas. Assuming the channel state information (CSI) is perfectly known, the asymptotic expression of spectral efficiency is obtained. Furthermore, when considering a crowded urban street with a lot of vehicles, the closed-form expression of spectral efficiency is obtained. Some novel and interesting conclusions are presented from the asymptotic and closed-form expressions: the movement of terminal hardly affects the spectral efficiency when CSI is perfectly known, and the spectral efficiency is determined by the vehicle amount, the antenna number, and the serving range of RSU. Numerical results verify the accuracy of the proposed expressions of spectral efficiency and give an insightful suggestion about system design: shrinking the serving expansion can help improve the system performance.

This paper investigates the performance of bidirectional dynamic networks (BDN) with massive multiple input multiple output (MIMO) systems. In BDN, dynamic allocation of the number of uplink and downlink remote radio heads (RRHs) is proposed, which offers a flexible solution to balance the data traffic asymmetry without requiring the time synchronization. Intuitively, the interference between the downlink and uplink RRHs is one of the main challenges in BDN. However, we prove that the massive MIMO strategy can effectively reduce a certain portion of the downlink-to-uplink interference. We derive the approximations of the achievable uplink and downlink rates using a maximum ratio transmission (MRT) precoder and a maximum ratio combination (MRC) receiver. Based on the asymptotic analysis, we exploit the power scaling laws that both the uplink and downlink power should scale down to 1/M (M is the antenna number) to ensure a desirable uplink or downlink rate. Furthermore, simulations show that BDN outperforms traditional time division duplex (TDD) systems in terms of the spectral efficiency.

With the development of mobile internet, mobile payment has become one of the most popular payment methods. In order to improve work efficiency, reduce labor costs and enhance user experience, the intelligent identification of bank cards is widely used in mobile payment. Conventional optical character recognition (OCR) technology has the problems of low recognition rate when dealing with bank card text with complex background. Thus, a bank card number identification method based on deep learning is proposed. Firstly, the data set is expanded. Then the CRNN algorithms is used and optimized to identify the card number. Some experimental results show that the method have a high recognition rate.

We consider the downlink multi-user multi-cell massive MIMO systems, and our system model accounts for channel estimation, pilot contamination, and uniformly random user location distribution. An equivalent channel model is investigated which indicates that the users in various cells come through the same ‘fast fading’ when they reuse the same training signal in uplink pilot transmission phase. Assuming that the number of antennas at base station (BS) is large, under the proposed equivalent channel model, we derive the asymptotic approximations of area spectral efficiency (ASE) with zero-forcing (ZF) and regularized zero-forcing (RZF) techniques which are proven to be accurate via simulation results. Furthermore, with a realistic power consumption model considering not only transmit power but also the precoding processing part, we analyze the performance of area energy efficiency (AEE). In particular, AEE shows quasiconcave with the number of antennas, and it indicates that we can obtain the optimal number of antennas with the existing convex methods.

This paper investigates the performance of massive\ud multiple input multiple output (MIMO) cellular systems with\ud pilot contamination. We derive a closed-form approximation of\ud the area spectral efficiency (ASE) for uplink multi-cell multiuser\ud massive MIMO systems by using an uniformly distributed\ud user location model. Considering a practical power consumption\ud model, we obtain the area energy efficiency (AEE) in a cell. The\ud theoretical results of ASE and AEE can be used to investigate\ud some system parameters of massive MIMO cellular system, such\ud as the number of antennas at the base station, the number of\ud users, and the number of symbols over which the channel remains\ud constant. We also study the optimal pilot-to-data power ratio\ud (PDPR) for both ASE maximization and AEE maximization.