The external power supply requirement of intelligent reflecting surfaces (IRS) restricts its deployment and application. To address this issue, self-sustainable IRS that can harvest energy from wireless signals significantly enhances the convenience and feasibility of IRS deployment. This paper investigates a novel IRS-aided simultaneous wireless information and power transfer (SWIPT) system, where IRS adopts power splitting (PS) protocol, time switching (TS) protocol or element splitting (ES) protocol to extract energy from the incident signals for sustaining its operation. Our aim is to maximize the weighted sum rate (WSR) while accounting for the constraints in terms of maximum transmit power, IRS reflection coefficients, the energy harvesting requirements of both users and the IRS. The WSR maximization problem is non-convex due to the coupling among variables. For the formulated problem, we propose an alternating optimization (AO) framework to transform it into several subproblems and solves them iteratively. Specifically, we utilize successive convex approximation (SCA) technique to tackle the non-convex optimization challenges in both transmit and reflective beamforming subproblems. Finally, simulation results demonstrate that employing a self-sustainable IRS together with the proposed algorithm yields a WSR improvement of 32%–60%.

The external power supply requirement of intelligent reflecting surfaces (IRS) restricts its deployment and application. To address this issue, self-sustainable IRS that can harvest energy from wireless signals significantly enhances the convenience and feasibility of IRS deployment. This paper investigates a novel IRS-aided simultaneous wireless information and power transfer (SWIPT) system, where the IRS adopts a power splitting (PS) protocol to extract energy from the incident signals for sustaining its operation. Our aim is to maximize the weighted sum rate (WSR) while considering the constraints in terms of maximum transmit power, IRS reflection, the harvesting energy requirements of users and IRS. The formulated problem of maximizing WSR is non-convex, which stems from the coupling among variables, making the problem complicated. To solve this problem, we propose an alternating optimization (AO) framework to transform the formulated problem into three subproblems, which enables an iterative solution approach. Specifically, we utilize successive convex approximation (SCA) technique to tackle the non-convex optimization challenges in both transmit and reflective beamforming subproblems. Simulation results demonstrate the efficacy of the proposed algorithm and substantiate the advantages of implementing self-sustained IRS for improving the WSR performance compared to other benchmark schemes.

In the domain of deep learning for visual tasks, ensemble models combine several less accurate models to form a more precise composite model, improving overall performance. Traditionally, majority voting and average probabilities have been the main decision-making techniques in ensemble learning, focusing only on the TOP1 Class of base models, hence overlooking other significant information. This article introduces a new algorithm, TOP2 hybrid decision (TOP2 HD), which enhances the TOP1 accuracy of the ensemble model. TOP2 HD categorizes base models into hierarchies based on their TOP1 Class and uses the TOP2 Class for ranking, leading to better performance. Extensive experiments across various models and datasets demonstrate that TOP2 HD not only surpasses traditional ensemble methods, such as majority voting, average probabilities, and stacking, but also exceeds many of the latest ensemble strategies in the image domain. In addition, our experiments revealed a functional relationship between the test accuracy of the ensemble model and the number of base models. This enables us to predict the upper limit of the ensemble model's performance using only a fraction of the models, providing a crucial reference for the performance after the deployment of the ensemble model.

Cell-free network and intelligent reflecting surface (IRS) are considered as promising technologies for future network capacity and coverage improvement. They offer advantages such as low cost, low energy consumption, and meeting the requirements of green communication. However, the fixed location of IRS limits the flexibility of the entire network. To address this issue, we propose a more comprehensive cell-free network that enhances network capacity and signal coverage by utilizing the reflected signals from an airborne IRS. Our objective is to maximize the weighted transmission rate for users by jointly optimizing the base station (BS) beamforming, passive beamforming of the IRS, and the location of the UAV. Owing to the non-convex and intricate nature of the problem, we decompose it into three subproblems, employing the principles of Lagrangian duality, multi-ratio fractional programming, and the successive convex approximation (SCA) technique for resolution. Simulation results demonstrate that the proposed scheme can significantly improve the weighted transmission rate and effectively enhance the network coverage compared to the benchmarks.

Currently, deep learning methods have been successfully applied to soil property content estimation from soil spectra due to their powerful feature extraction capability. In practical production, it is necessary to estimate the contents of multiple soil properties simultaneously. The accuracy of such estimation heavily depends on the ability of the algorithm to balance multiple estimation tasks. In this letter, a multi-task learning network combining attention mechanism and loss-weight balancing approach based on feature correlation is proposed. First, a parameter-sharing module of a three-layer convolutional neural network (CNN) is constructed. Second, an independent channel importance recalculation module is constructed for each estimation task, which consists of an efficient channel attention (ECA) module. Finally, the features extracted from these two modules are concatenated, and a two-layer CNN is constructed to further extract features for estimating each soil component. Moreover, an improved loss-weight uncertainty technique based on the correlation between soil spectra and property contents is proposed to reconcile the learning effects of multiple estimation tasks. The experimental results on two soil datasets, LUCAS (Land Use/Land Cover Area Frame Survey) 2009 and AfSIS (Africa Soil Information Service), show that this method provides competitive accuracy compared with several state-of-art methods.

The fifth-generation Internet of Things (5G-IoT) requires more capable channel coding methods to obtain low-latency and high-reliability communication systems. Hence, an improved ordered tree-based single-parity-check (OT-SPC) list decoder for polar code is designed in this article to reduce the complexity and keep the reliability. The proposed method introduces the ordered error pattern set of the single-parity-check (SPC) list decoder, which collects all the necessary error patterns. We first define a relationship between these patterns and present the concepts of the necessary error pattern and the ordered error pattern set. Then, we adapt the ordered tree data structure and the asynchronous tournament sorters to relieve the complexity increase of the ordered error pattern set’s introduction. The reliability relationships of different patterns are stored in an ordered tree, and the tournament sorting algorithm is modified separately for the log-likelihood ratio values of the received sequence and the path metrics of the decoding paths. In this scheme, the candidate codewords of the SPC list decoder are constructed in the descending order of their probability, which significantly reduces the calculation and comparison consumption. Eventually, the results show that the OT-SPC list decoder obtains a superior performance in decoding throughput and latency while preserving the error–correction performance.

At present, deep learning method relies on its strong feature extraction ability has been successfully applied to the estimation of soil organic carbon (SOC) content with hyperspectral data. However, due to the high dimensionality of hyperspectral data and equal treatment of all bands, the performance of these methods is hampered by learning features from useless bands. To address this issue, in this letter, attention mechanism is combined with convolutional neural network (CNN) to assign different weights to different bands of the hyperspectral data. This method constructs a three-layer CNN with a squeeze-and-excitation module at the front of it. Then, five attention-based CNNs are combined to establish an ensemble regression system with diversity. The inputs of each branch in this system are the original hyperspectral data and its transformed data. Moreover, an improved label distribution smoothing technique is proposed to address the problem of imbalanced samples. The experimental results on three soil datasets, LUCAS (Land Use/Land Cover Area Frame Survey) 2009, LUCAS2015 and AfSIS (Africa Soil Information Service), show that this method obtains good estimation performance compared with several state-of-art methods, especially in the areas with high SOC content which has small sample sizes.

Since the content of soil organic carbon (SOC) is significantly correlated with the soil reflection spectrum, hyperspectral remote sensing technology provides an effective new choice for the estimation of soil properties. At present, deep learning method relies on its strong feature extraction ability and has been successfully applied to the field of data analysis. This paper attempts to apply the deep learning method to the estimation of SOC content and proposes a method combining the convolutional neural networks (CNN) and quantile regression (QR). This method constructs a three-layer CNN and adjusts the network structure with the idea of QR. The experimental results are presented for two soil datasets, LUCAS (Land Use/Land Cover Area Frame Survey) and AfSIS (Africa Soil Information Service), and compared with several advanced deep learning and traditional machine learning regression models. The experimental results show that this method performs well in estimation.

The security of wireless information transmission in large‐scale multi‐input and multioutput (MIMO) is the focus of research in wireless communication. Recently, a new artificial noise—SCO‐AN which shows no orthogonality to the channel, is proposed to overcome the shortcomings of traditional artificial noise. In the previous research, the optimization function of SCO‐AN is not convex, and its extremum cannot be obtained. Usually, nonconvex optimization algorithms or iterative relaxation algorithms are used to get the maximum value of the optimization objective function. Nonconvex optimization algorithms or iterative relaxation algorithms are greatly affected by the initial value, and the extremum cannot be obtained by a nonconvex optimization algorithm or iterative relaxation algorithm. In this paper, we creatively apply the strong law of large numbers to obtain the optimal value of the optimization function of SCO‐AN under the condition of large‐scale MIMO: the strong law of large numbers is applied to obtain the ergodic lower bound (ELB) expression of SC for SCO‐AN. The power allocation (PA) problem of the SCO‐AN system is discussed. We use a statistical method to get the formula for calculating the optimal power distribution coefficient of the SCO‐AN system. The transmitter can use the optimal power ratio of PA to distribute the transmitted power without using the PA algorithm. The effect of imperfect channel state information is discussed. Through simulation, we found that more power should be generated for SCO‐AN if the channel estimation is imperfect and the proposed method can achieve better security performance in the large‐scale MIMO system.

Ultra-wideband (UWB) technology is a potential candidate solution for Wireless Sensor Network (WSN) and Internet of Things (IoT) applications due to the advantages of high-speed communication and accurate positioning. Channel estimation plays a key role in UWB systems, especially in low signal-to-noise ratio (SNR) environments. In this paper, a noise-resistance channel estimation method based on compressed sensing is proposed for accurate estimation of UWB channels contaminated by noise. The proposed algorithm based on compressed sensing relies on the fact that the multipath signals of UWB have a sparse representation in the time domain or other domains. However, the reconstruction of multipath signals is difficult owing to the additive noise and noise-folding effect. In order to improve the noise robustness, UWB multipath signals are sparse representation by using the orthogonal basis of an eigen-based dictionary. Three approaches are applied to the proposed noise-resistance algorithm, including average pilot noise reduction, denoising whitening measurement, and sparsity adaptive matching pursuit (SAMP) algorithm. The improved relative residue is proposed to calculate the iteration termination condition of SAMP. Simulation results demonstrate that the algorithm provides good noise-resistance performances in terms of normalized mean square error (NMSE) of channel estimation and the bit error rate of Rake receivers, especially in low SNR environments.

In cognitive radio network, the licensed spectrum for the primary user can be accessed in an opportunistic manner by secondary user, or unlicensed user. As a key technology of cognitive radio, spectrum sensing has an irreplaceable position. In this paper, we proposed a parallel CNN-LSTM network based deep learning algorithms for spectrum sensing. As much modulated signals and noise data as possible are generated to train the model to accommodate detection of multiple types signal. Various experiments are performed to prove the effectiveness of proposed method, and requiring no prior knowledge about the information of licensed user or channel state. The simulation results show that the model can detect multiple modulation types under a large scale of SNRs, especially in low SNR.

The research for co-channel multi-signal modulation classification has become urgent with the increasing shortage of spectral bandwidth. Single-signal modulation classification methods which have been widely studied are not applicable for co-channel multi-signal modulation classification problem. In this paper, we developed a method for co-channel multi-signal modulation classification based on Convolution Neural Network(CNN). The proposed method can identify 31 mixed signals from 5 modulation types. The proposed method are also found to be robust to the changes of SNR from 0dB to 15dB. The experiments are performed to prove the effectiveness of the proposed method.

Modern intelligent transport systems focus on the integration of multiple sensors to obtain hybrid navigation schemes. A key issue of a hybrid scheme is distribution of the information sharing coefficients (ISCs) of subsystems and the fusion of parallel multiple observations of navigation sensors. Recently, deep learning methods, particularly convolutional neural networks (CNNs), have achieved great success in image processing tasks. However, there has been limited work in using deep learning for multisensor-based integrated navigation solutions. In this letter, we propose an ensemble learner-based classification and information fusion method, in which estimation error covariance matrices provided by local adaptive filters are used as input for the classifier, and the triple numbers of ISCs are determined by the proposed scheme. The results validate the effectiveness of the proposed scheme, in which the adequately trained ensemble learner can detect the degradation of a subsystem that may suffer atypical observations or faults and consequently can adjust the corresponding ISC in real time.

Automatic modulation classification (AMC) plays an important role in cognitive radio and is widely studied recent years. However, most existing AMC schemes must be deployed under their training SNRs, which makes them highly dependent on the accuracy of channel estimation. The classifiers may need to be re-trained to fit the varying channel condition. To address this problem, a feature combination method aiming to find noise-robust features under fading channels is proposed in this paper. Stacked auto encoder is deployed to explore robust features from an extracted feature set, and these new features is then used to train a support vector machine (SVM). Numerical results shows that the generalization ability of SVMs trained with new features can be significantly improved; therefore the method is robust to SNR variation.

Various vehicle applications in the future will require reliable and accurate vehicle positioning techniques. Nowadays, hybrid schemes combining multiple sensors have been promising solutions for high precision positioning. However, positioning error can be remarkably affected by the temporal alignment and fusion algorithms in practice. In this paper, we propose a decentralized fusion structure containing an inertial navigation system (INS), a GPS receiver, a RFID reader, and an odometer. The update rates of the sensors are different, and the INS/GPS integration presents severe performance degradation in urban area. To achieve an effective alignment and fusion of the sensors, we propose a concept of confidence field to indicate the confidence levels of subsystems for changing driving environments. A confidence field-based alignment and fusion algorithm and its simplification are proposed when we use the weighted least squares curve method. Time biases of the sensors are also considered in local adaptive filters. Simulation results demonstrates the performance of the proposed scheme with the proposed algorithms, especially in GPS- denied environments.

The performance of Global Satellite Positioning System / Inertial Navigation System (GPS/INS) integrated navigation system based on Kalman Filter (KF) is greatly influenced by measurement information related to GPS. However, it can be unreliable: it can be lost and the statistical characteristics of the measurement noise can change. Thus, the performance of navigation will get worse. Therefore, a filter algorithm for the integrated navigation system based on the Interacting Multiple Model-Adaptive Filter (IMM-AF) is proposed in this paper. Two measurement noise models for small Gaussian noise and non-small Gaussian noise are designed respectively to be applied to the algorithm; one step prediction algorithm for the case of GPS signal loss is also combined. The results of the experiment of the integrated navigation system of mobile robot show that, compared with KF or IMM, IMM-AF algorithm presents higher accuracy and better robustness, with almost the same update time.