An intelligent reflecting surface (IRS) is a programmable device that can be used to control electromagnetic waves propagation by changing the electric and magnetic properties of its surface. Therefore, IRS is considered a smart technology for the sixth generation (6G) of communication networks. In addition, machine learning (ML) techniques are now widely adopted in wireless communication as the computation power of devices has increased. As it is an emerging topic, we provide a comprehensive overview of the state-of-the-art on ML, especially on deep learning (DL)-based IRS-enhanced communication. We focus on their operating principles, channel estimation (CE), and the applications of machine learning to IRS-enhanced wireless networks. In addition, we systematically survey existing designs for IRS-enhanced wireless networks. Furthermore, we identify major issues and research opportunities associated with the integration of IRS and other emerging technologies for applications to next-generation wireless communication.

Visible light communication (VLC) has contributed new unused spectrum in addition to the traditional radio frequency communication and can play a significant role in wireless communication. The adaptation of VLC technology enhances wireless connectivity both in indoor and outdoor environments. Multiple-input multiple-output (MIMO) communication has been an efficient technique for increasing wireless communications system capacity and performance. With the advantages of MIMO techniques, VLC can achieve an additional degree of freedom. In this paper, we systematically perform a survey of the existing work based on MIMO VLC. We categorize the types of different MIMO techniques, and a brief description is given. Different problem-solving approaches are given in the subsequent sections. In addition, machine learning approaches are also discussed in sufficient detail. Finally, we identify the future study direction for MIMO-based communication in VLC.

Deep learning (DL) techniques can significantly improve successive interference cancellation (SIC) performance for the non-orthogonal multiple access (NOMA) system. The NOMA-orthogonal frequency division multiplexing (OFDM) system is considered in this paper to develop a hybrid deep neural network (HyDNN) model for multiuser uplink channel estimation (CE) and signal detection (SD). The proposed HyDNN uses a combination of a bi-directional long short-term memory (BiLSTM) network and a one-dimensional convolutional neural network (1D-CNN) to optimize errors in the system. The extraction of input signal characteristics from OFDM is carried out using the 1D-CNN model and fed into the time series BiLSTM network to infer the signal at the receiver terminal. The HyDNN model learns through the simulated channel data during offline training. To optimize the loss during learning the model the Adam optimizer is utilized. After successful training, the transmitted symbols in the online deployment are instantly recovered with optimal prediction rates by using the proposed HyDNN model. In comparison to the traditional CE and SD method for the NOMA scheme and other existing DL models, the proposed technique demonstrates satisfactory performance enhancements. In addition, the simulation outcomes show robustness with different training parameters such as minibatch sizes and learning rates.

In this paper, we propose the adaptive modulation (AM) model based on machine learning for multiple input multiple output (MIMO) - orthogonal frequency division multiplexing (OFDM) system. Since 5G new radio (NR) system can be used in a large variety of internet of things (IoT) fields than the conventional systems, the AM scheme that adjusts data rate and reliability according to channel condition can be effectively utilized. The conventional AM schemes are implemented by defining modulation schemes to be used according to each channel condition as table in advance. However, since the rulebased AM can not analyze the communication performance according to the correlation of channels between antennas and the number of transmission modes is exponentially increased according to the number of available modulation schemes and antennas, it is not suitable for 5G NR system. The learning of proposed AM model is based on the generated training signal by the extracted features from received signal and assigned label through performance analysis for signal detection. We focus on the application of deep neural network (DNN) for AM and covers the precedence method of principal component analysis (PCA) to improve the performance of model. Simulations on the classification of optimal transmission mode for MIMOOFDM signal demonstrate that the proposed model supports the adaptability according to condition of complex MIMO channel.

In this paper, a complexity-effective integer carrier frequency offset (IFO) and primary synchronization signal (PSS) detection scheme is proposed for 5G new radio systems. To estimate the IFO without any prior information about the PSS, the proposed detector uses the fact that three PSS sequences are cyclic-shifted versions of each other, which enables sequential detection of IFO and PSS. Both theoretical and simulation results are presented to confirm the usefulness of the proposed synchronization scheme. It is shown that the proposed approach achieves a considerable complexity reduction while suffering from only slight degradation in the detection performance, compared to the conventional method.

Non-orthogonal multiple access (NOMA) has great potential to implement the fifth-generation (5G) requirements of wireless communication. For a NOMA traditional detection method, successive interference cancellation (SIC) plays a vital role at the receiver side for both uplink and downlink transmission. Due to the complex multipath channel environment and prorogation of error problems, the traditional SIC method has a limited performance. To overcome the limitation of traditional detection methods, the deep-learning method has an advantage for the highly efficient tool. In this paper, a deep neural network which has bi-directional long short-term memory (Bi-LSTM) for multiuser uplink channel estimation (CE) and signal detection of the originally transmitted signal is proposed. Unlike the traditional CE schemes, the proposed Bi-LSTM model can directly recover multiuser transmission signals suffering from channel distortion. In the offline training stage, the Bi-LTSM model is trained using simulation data based on channel statistics. Then, the trained model is used to recover the transmitted symbols in the online deployment stage. In the simulation results, the performance of the proposed model is compared with the convolutional neural network model and traditional CE schemes such as MMSE and LS. It is shown that the proposed method provides feasible improvements in performance in terms of symbol-error rate and signal-to-noise ratio, making it suitable for 5G wireless communication and beyond.