Deep-Learning-Based Spatial–Temporal Channel Prediction for Smart High-Speed Railway Communication Networks

A Multifrequency Seamless Dual-Link Handover Scheme Based on Beamforming for High-Speed Railway

“…It must be mentioned that the research of future smart HSR communication based on machine learning (ML) has been paid more attention and made benefcial progress. Literature [22] investigates the spatial-temporal prediction of channel state information and channel statistical characteristics based on deep-learning (DL) for the future smart HSR communication, a novel spatial-temporal channel prediction model that combines the convolutional neural network and convolutional long short-term memory is proposed by exploiting the temporal and spatial correlations of massive MIMO channel in HSR. Literature [23] investigates the channel multipath components (MPCs) clustering based on ML and analyzes the cluster characteristics in typical HSR scenarios, a variational Bayesian Gaussian mixture model-based algorithm is introduced to achieve the space-time clustering of MPCs.…”

Learning-Based Joint Channel Prediction and Antenna Selection for Massive MIMO with Partial CSI

“…Because there are only N f ≤ N t RF chains, the BS must select a best subset of N f antennas for downlink data transmission. For the AS problem, a common objective is to optimize a generic objective function F k (a a a) while obeying Solutions to problem (3) have been well studied in the literature [4], [6], [7], [11]- [13] under the complete CSI assumption, meaning that the wireless channel is fully observable. Since the number of RF chains is less than the number of transmit antennas at BS, obtaining complete CSI requires τ f ull csi = N u (⌊ Nt N f ⌋ + 1) c.u.…”

“…[8]- [10]. From this perspective, many channel prediction methods have recently been investigated in [11]- [13]. A machine learning (ML)-based time-division duplex scheme was proposed in [11], where full CSI is obtained by leveraging the temporal channel correlation that is applied to both low and high mobility scenarios.…”

“…A machine learning (ML)-based time-division duplex scheme was proposed in [11], where full CSI is obtained by leveraging the temporal channel correlation that is applied to both low and high mobility scenarios. To predict channels smart high-speed railway communication networks, the authors of [13] proposed a channel prediction scheme based on convolutional neural network (CNN) and long short-term memory network (LSTM) which predicts full CSI in a multi-step ahead manner by exploiting the channel correlations. Nevertheless, the prediction methods in [11]- [13] require the fully observed channels history and are not applicable to the case of partial observation, especially when the number of antennas exceeds the number of RF chains.…”

“…To predict channels smart high-speed railway communication networks, the authors of [13] proposed a channel prediction scheme based on convolutional neural network (CNN) and long short-term memory network (LSTM) which predicts full CSI in a multi-step ahead manner by exploiting the channel correlations. Nevertheless, the prediction methods in [11]- [13] require the fully observed channels history and are not applicable to the case of partial observation, especially when the number of antennas exceeds the number of RF chains. Recently, AS with partial CSI has been proposed in [14] based on multi-armed bandit (MAB) and Thompson sampling technique to reduce the channel estimation overhead.…”