Millimeter-wave massive multiple-input multiple-output (MIMO) employs a lens antenna array for minimizing the radio frequency (RF) chains count, yet it remains as a challenge since the RF chain count is lesser when compared to the antennas. The beamspace channel estimation is defined as a sparse signal recovery problem through the exploitation of sparsity of beamspace channels. In the case of multi-user millimeter wave (mmWave) MIMO-OFDM systems, the critical task to lessen the cost as well as the complexity is the hybrid precoding attaining an adequate sum-rate. The conventional approaches related to the hybrid precoding works on the basis of the greedy or optimization techniques. These techniques suffer from higher complexity or contain sub-optimum performance. The main intent of this paper is to plan for the adaptive deep learning strategy for sparse channel estimation and hybrid precoding in millimeter-wave massive MIMO-OFDM communication system. The proposed methodology covers two main phases (a) uplink channel estimation, and (b) hybrid precoding for downlink data transmission. In the first phase, an adaptive deep neural network (ADNN) is used for performing the uplink channel estimation. Here, the benchmark dataset is used for training the information to be predicted at base station regarding the channel. Here, the ADNN-based channel estimation covers both channel amplitude estimation and channel reconstruction. Once the channel reconstruction is done, hybrid pre-coding performs the downlink data transmission that involves both digital and analog pre-coders. Here, the adaptive long short term memory (ALSTM) is used for performing the hybrid pre-coding, in which the estimated channel vectors is used for training. For both sparse channel estimation and hybrid-precoding, the trial-based red deer algorithm (T-RDA) is used for improvising the ADNN and ALSTM, in which it optimizes or tunes the training hidden neurons. The main objective of tuning the hidden neurons in both phases of T-RDA is to maximize the accuracy. When SNR equals 15 dB, the NMSE of the T-RDA-ADNN is 40%, 25%, 37.5%, and 34.78% improved than RDA-ADNN, SFO-ADNN, GWO-ADNN, and PSO-ADNN for channel estimation. Similarly, at SNR 20 dB, the spectral efficiency of T-RDA-ADNN is 37.5%, 43.48%, 22.22%, and 94.12% advanced than RDA-ADNN, SFO-ADNN, GWO-ADNN, and PSO-ADNN for hybrid precoding. Thus, the simulation outcomes show that the developed sparse channel estimation and hybrid pre-coding in mmWave massive MIMO-OFDM communication system has attained lesser error and high spectral efficiency when differentiated over the existing approaches.

Adaptive Deep Learning Strategy with Red Deer Algorithm for Sparse Channel Estimation and Hybrid Precoding in Millimeter Wave Massive MIMO-OFDM systems

Multiple Input Multiple Output-Orthogonal Frequency Division Multiplexing (MIMO-OFDM) system is a promising technology that provides high capacity and high data rate transmission in 5G and beyond. Presence of a large number of antennas in a MIMO-OFDM system increases the overhead on pilot symbols used for channel estimation. Our proposed work exploits the channel sparsity to design a pilot based compressed sensing method for channel estimation and integrates a Bi-LSTM approach for symbol detection for improved performance. Additionally, we optimize the pilot symbols to minimize the error value to be effectively within the total power constraint. We evaluate the performance of our proposed system using mean square error (MSE) with least minimum mean square estimate (LMMSE) as the benchmark. We demonstrate the design and evaluation of a scalable and efficient approach to joint channel estimation and symbol detection in a MIMO-OFDM system using a fewer number of pilot symbols. 

Joint Channel Estimation and Symbol Detection in MIMO-OFDM Systems: A Deep Learning Approach using Bi-LSTM

Despite high costs and lengthy deployments, satellite communications have traditionally been used to provide coverage in remote areas. However, given the fact that there is no radio infrastructure available in these areas, Near Vertical Incidence Skywave (NVIS) technology has positioned itself as an attractive alternative to communicate with low-power nodes in remote areas. This type of communication works in the HF frequency range complying with STANAG and MIL-STD standards, which define a physical layer for scenarios that differ from NVIS and low-power communication. The purpose of this paper was to present the definition of a new communication physical layer based on single-carrier frequency-domain equalization (SC-FDE) based on these standards but adapted to the ionospheric communication channel. This physical layer was compared to an OFDM-based layer from a previous study. The experiments performed show that this new approach achieves better results than OFDM in terms of a higher signal quality with a higher specific BER probability. Finally, this layer was also used in the theoretical design of an NVIS gateway to link sensor network devices spanning large-scale remote areas in a secure manner in the context of ubiquitous sensor networks (USN).

SC-FDE Layer for Sensor Networks in Remote Areas Using NVIS Communications

Abstract At cellular wireless communication systems, channel estimation (CE) is one of the key techniques that are used in Orthogonal Frequency Division Multiplexing modulation (OFDM). The most common methods are Decision‐Directed Channel Estimation, Pilot-Assisted Channel Estimation (PACE) and blind channel estimation. Among them, PACE is commonly used and has a steadier performance. Applying deep learning (DL) methods in CE is getting increasing interest of researchers during the past 3 years. The main objective of this paper is to assess the efficiency of DL-based CE compared to the conventional PACE techniques, such as least-square (LS) and minimum mean-square error (MMSE) estimators. A simulation environment to evaluate OFDM performance at different channel models has been used. A DL process that estimates the channel from training data is also employed to get the estimated impulse response of the channel. Two channel models have been used in the comparison: Tapped Delay Line and Clustered Delay Line channel models. The performance is evaluated under different parameters including number of pilots (64 pilots or 8 pilots), number of subcarriers (64), the length of cyclic prefix (16 or 0 samples) and carrier frequency (4 GHz) through computer simulation using MATLAB. From the simulation results, the trained DL estimator provides better results in estimating the channel and detecting the transmitted symbols compared to LS and MMSE estimators although, the complexity of the proposed LSTM estimator exceeds the equivalent LS estimator. Furthermore, the DL estimator also demonstrates its effectiveness with various pilot densities and with different cyclic prefix periods. 

/pdf/deep-learning-channel-estimation-for-ofdm-5g-systems-with-3g7ckdnc.pdf

Deep Learning Channel Estimation for OFDM 5G Systems with Different Channel Models

This paper presents our initial results in assessing the efficiency of deep learning-based channel estimation compared to the conventional Pilot-Assisted Channel Estimation (PACE) techniques, such as least-square (LS) and minimum mean-square error (MMSE) estimators. A simulation environment to evaluate OFDM performance at different channel models has been used. A DL process that estimates the channel from training data is also employed to get estimated channel impulse response. Two channel models have been used in the comparison: Clustered Delay Line (CDL) and Nakagami-m fading channel models. The performance is evaluated under different parameters including number of pilots, number of subcarriers, the length of cyclic prefix and carrier frequency through computer simulation using MATLAB. From the simulation results, the trained DL estimator provides better results in estimating the channel and detecting the transmitted symbols compared to LS and MMSE estimators at a remarkably less complexity. Furthermore, the DL estimator also demonstrates its effectiveness with various pilot densities and with different cyclic prefix periods. 

The long distance high frequency (HF) communication suffers from time varying multipath fading. Multiple-Input Multiple-Output (MIMO) and Single-Carrier Frequency Domain Equalization (SC-FDE) have been introduced to HF communication system to combat fading. Because accurate channel estimation is essential for system operation, an architecture of signal flow chart has been designed for the HF MIMO SC-FDE system. In the architecture, an online Long Short-Term Memory (LSTM) estimator is proposed. Different from the channel responses time series created by the LSTM training and prediction loop, a corrected channel responses that are obtained through the received data symbols and the restored transmitting data symbols make up the channel responses time series. In order to evaluate the performance of channel estimators, a simulation system has been built. The uncorrelated and correlated channels are simulated referring to International Telecommunication Union (ITU)-R F.1487 standard and Kronecker model. The simulation results demonstrate that the online LSTM estimator outperforms Least Square (LS) and Recursive Least Square (RLS) estimators in terms of Bits Error Rate (BER) and Mean Square Error (MSE). The online LSTM estimator is capable of tracking the time varying HF MIMO channels. It has potentiality in actual long distance HF communication.

/pdf/online-lstm-based-channel-estimation-for-hf-mimo-sc-fde-4i2utj2ncb.pdf

Rui Yang

Papers

Online LSTM-Based Channel Estimation for HF MIMO SC-FDE System