This two-part paper investigates the application of artificial intelligence (AI) and, in particular, machine learning (ML) to the study of wireless propagation channels. In Part I of this article, we introduced AI and ML and provided a comprehensive survey on ML-enabled channel characterization and antenna-channel optimization, and in this part (Part II), we review the state-of-the-art literature on scenario identification and channel modeling here. In particular, the key ideas of ML for scenario identification and channel modeling/prediction are presented, and the widely used ML methods for propagation scenario identification and channel modeling and prediction are analyzed and compared. Based on the state of the art, the future challenges of AI-/ML-based channel data processing techniques are given as well.

To provide higher data rates, as well as better coverage, cost efficiency, security, adaptability, and scalability, the 5G and beyond 5G networks are developed with various artificial intelligence (AI) techniques. In this two-part article, we investigate the application of AI and, in particular, machine learning (ML) to the study of wireless propagation channels. It first provides a comprehensive overview of ML for channel characterization and ML-based antenna–channel optimization in this first part, and then, it gives a state-of-the-art literature review of channel scenario identification and channel modeling in Part II. Fundamental results and key concepts of ML for communication networks are presented, and widely used ML methods for channel data processing, propagation channel estimation, and characterization are analyzed and compared. A discussion of challenges and future research directions for ML-enabled next-generation networks of the topics covered in this part rounds off this article.

Vehicle-to-vehicle (V2V) communications have received a lot of attention as it can significantly improve efficiency and safety of road traffic. In general, V2V scenarios have some special features such as high mobility of transceivers, low antenna height, and small communication ranges, which significantly affect the propagation characteristics. Therefore, an accurate channel model is required to characterize V2V propagation channel. However, some important features of V2V channels have not been well characterized. For example, most existing V2V channel models tend to only consider the distribution of multipath components (MPCs) in horizontal dimension, whereas ignoring vertical dimension, which is inconsistent with the distribution of MPCs in the actual channel. Moreover, the dynamic clusters of MPCs have not been well modeled in the existing V2V models. Therefore, in order to better model the V2V channel, a cluster-based three-dimensional (3D) channel model is proposed in this paper, which is based on the measurements conducted at 5.9 GHz in urban and suburban scenarios. In the proposed model, the distribution of MPCs clusters in both horizontal and vertical dimensions is considered. The Space-Alternating Generalized Expectation-maximization (SAGE) algorithm is used to extract MPCs, and the clustering and tracking algorithms are used to identify and track the clusters of dynamic MPCs. In the proposed model, all MPC clusters are divided into two categories: global-clusters and scatterer-clusters, and the distribution of the two clusters are characterized by a series of inter and intra-cluster parameters. It is found that both the azimuth spread and the elevation spread follow the lognormal distribution. In addition, the power of MPCs within a cluster has the truncated-Gaussian distribution, whereas the angle of MPCs within a cluster has the Laplacian distribution. Finally, the accuracy of the model is verified by comparing the measurements and simulations.

Obtaining angle-of-arrival (AOA) information is of great significance to improve the performance of communication systems. Real-time AOA recognition can reduce the complexity of beamforming design for massive multiple-input multiple-output (MIMO) systems, and can be used to construct an outer precoder to optimize the system and rate. However, for vehicular communications, AOA will change with environment and positions of vehicles, and it is difficult to obtain accurate AOA in real-time. Therefore, a fast AOA recognition method is needed to adapt to the rapid changes of channels. The traditional spectral- or parametric-based AOA estimation methods are difficult to obtain real-time AOA information because of the relatively high computational complexity. In order to solve this problem, this paper proposes a machine-learning-based fast AOA recognition approach. The proposed method includes off-line training and on-line estimation processes. In the off-line training process, an estimation model is obtained by using the support vector machine (SVM) based on a large number of actual measurement data in vehicular scenarios. Then, in the on-line estimation process, the obtained model is used to realize fast AOA recognition according to the channel snapshots collected by antenna array. Furthermore, the performance is verified under the different conditions of SVM parameters, training features, antenna numbers, and training data sizes. The experimental results show that the proposed method has satisfactory accuracy in real-time AOA recognition, and the optimal configuration and implementation scheme are also discussed.