Identifying line-of-sight (LOS) and non-LOS (NLOS) channel conditions can improve the performance of many wireless applications, such as signal strength-based localization algorithms. For this purpose, channel state information (CSI) obtained by commodity IEEE 802.11n devices can be used, because it contains information about channel impulse response (CIR). However, because of the limited sampling rate of the devices, a high-resolution CIR is not available, and it is difficult to detect the existence of an LOS path from a single CSI measurement, but it can be inferred from the variation pattern of CSI over time. To this end, we propose a recurrent neural network (RNN) model, which takes a series of CSI to identify the corresponding channel condition. We collect numerous measurement data under an indoor office environment, train the proposed RNN model, and compare the performance with those of existing schemes that use handcrafted features. The proposed method efficiently learns a non-linear relationship between input and output, and thus, yields high accuracy even for data obtained in a very short period.

Vehicle-to-everything communication system is a strong candidate for improving the driving experience and automotive safety by linking vehicles to wireless networks. To take advantage of the full benefits of vehicle connectivity, it is essential to ensure a stable network connection between roadside unit (RSU) and fast-moving vehicles. Based on the extended Kalman filter (EKF), we develop a vehicle tracking algorithm to enable reliable radio connections. For the vehicle tracking algorithm, we focus on estimating the rapid changes in the beam direction of a high-mobility vehicle while reducing the feedback overhead. Furthermore, we design a beamforming codebook that considers the road layout and RSU. By leveraging the proposed beamforming codebook, vehicles on the road can expect a service quality similar to that of conventional cellular services. Finally, a beamformer selection algorithm is developed to secure sufficient gain for the system's link budget. Numerical results verify that the EKF-based vehicle tracking algorithm and the proposed beamforming structure are more suitable for vehicle-to-infrastructure networks compared to existing schemes.

Detecting, measuring, and classifying partial discharges (PDs) are important tasks for assessing the condition of insulation systems used in different electrical equipment. Owing to the implementation of the phase-resolved PD (PRPD) as a sequence input, an existing method that processes sequential data, e.g., the recurrent neural network, using a long short-term memory (LSTM) has been applied for fault classification. However, the model performance is not further improved because of the lack of supporting parallel computation and the inability to recognize the relevance of all inputs. To overcome these two drawbacks, we propose a novel deep-learning model in this study based on a self-attention mechanism to classify the PD patterns in a gas-insulated switchgear (GIS). The proposed model uses a self-attention block that offers the advantages of simultaneous computation and selective focusing on parts of the PRPD signals and a classification block to finally classify faults in the GIS. Moreover, the combination of LSTM and self-attention is considered for comparison purposes. The experimental results show that the proposed method achieves performance superiority compared with the previous neural networks, whereas the model complexity is significantly reduced.

This letter considers a wireless powered communication network (WPCN), where an energy-constrained device directly uses harvested energy from a power transfer source to transmit independent signals to multiple Internet of Thing (IoT) users using orthogonal frequency division multiple access (OFDMA). Our goal is to maximize the system energy efficiency (EE) by jointly optimizing the duration of energy harvesting (EH), subcarrier and power allocation. The formulated problem is a mixed integer nonlinear programming (MINLP) problem due to the presence of binary assignment variables, and thus it is very challenging to solve it directly. By leveraging Dinkelbach method, a very efficient iterative algorithm with closed-form solutions in each iteration is developed, where its convergence is guaranteed. Numerical results show that the proposed algorithm obtains a fast convergence and outperforms baseline algorithms. Notably, they also reveal that the power source should transmit its maximum allowable power to obtain the optimal EE performance.

A relay selection method is proposed for physical-layer security in multi-hop decode-and-forward (DF) relaying systems. In the proposed method, cooperative relays are selected to maximize the achievable secrecy rates under DF-relaying constraints by the classification method. Artificial neural networks (ANNs), which are used for machine learning, are applied to classify the set of cooperative relays based on the channel state information of all nodes. Simulation results show that the proposed method can achieve near-optimal performance for an exhaustive search method for all combinations of relay selection, while computation time are reduced significantly. Furthermore, the proposed method outperforms the best relay selection method, in which the best relay in terms of secrecy performance is selected among active ones.