Communication technology has brought unprecedented convenience in information transmission and acquisition. Symmetry plays an important role in the field of communication engineering. Channel symmetry between uplink and downlink as well as frequency division duplexing communication systems have been fully utilized to date. Symmetric algorithm, topology and connection have been widely applied in communication networks. With the rapid increase in communication requirements, there are some challenging issues for symmetric FDD communications, which is the case of this article. Large-scale and small-scale channel characteristics are two important aspects of channel modeling. The former is closely related to the interval of the deployment spacing of base stations along the railway. When base stations are densely deployed, although the signal coverage level (SS-RSRP) can be guaranteed, it can also lead to severe interference between 5G-R cells, as well as high costs for base station equipment, towers, station buildings, supporting transmission and power equipment, infrastructure, land acquisition, engineering construction and other issues. When the deployment distance of the base station is too large, it will cause the received signal level of the train to be low when it reaches the edge of the cell, resulting in the inability to guarantee the signal-to-noise ratio (SS-SINR), leading to a low transmission rate and the poor connection reliability of 5G-R carrying services. As for the small-scale characteristics, the high-speed movement of trains leads to small-scale channel propagation problems such as the rapid fading of wireless channels and severe Doppler spread, resulting in a significant decline in the performance of railway 5G-R railway dedicated communication systems, directly affecting the transmission rate and reliability of railway communication applications carried by 5G-R.
Railway radio communication, as the foundation of promoting efficiency as well as reliability of railway transportation, reliably carries key services such as dispatching communication, train control information, maintenance communication, etc, and is closely related to the safety of passengers’ lives. Since frequency resource is the physical base and premise of railway radio communication, to allocate and protect dedicated frequency resources to railway radio communication system is a common consensus, from the point of view of the International Union of Railway as well as many non-European countries.
The leading global system of railway dedicated mobile communication (GSM-R), using railway dedicated frequency mentioned above, has been a great success, which has covered 163,000 km of railway lines in Europe and 90,000 km of railway lines in China. However, GSM-R, as a narrow band system, is unable to meet the constantly evolving demand for train-ground transmission bandwidth. Future Railway Mobile Communication System (FRMCS) is a project of International Railway Union (UIC) which aims to research and develop the successor of GSM-R, and 5G-R was chosen by FRMCS as the future mobile communication standard for railway; in the meanwhile, however, the 5G band 1900–1910 MHz, which is defined as n101 by 3GPP, is allocated to the 5G-R system in Europe. China Railway also adopted 5G-R for the new generation mobile communication, and in September 2023, the Ministry of Industry and Information Technology of China (MIIT) issued a document permitting China Railway to conduct tests on a test line of loop railway located in the northeast part of Beijing, China, using a frequency of 2100 MHz (1965–1975 MHz/2155–2165 MHz, FDD), defined as n1 by 3GPP, which is relatively close to that of Europe. According to a survey conducted by GSMA, the 2100 MHz frequency band ranks among the top three most frequently adopted frequency bands, second only to the 3500 MHz and 700 MHz frequency bands in terms of 5G deployment, as seen in
Figure 1.
The networking of a railway mainline mobile communication system is narrow-strip-shaped, which means that base stations are deployed along the railway line. Since railways are ramified all over the country, operating scenarios have to be considered, including viaducts, cuttings, stations, hilly terrain, as well as open space [
1]. Apart from special propagation scenarios, a high moving speed over 350 km/h, a higher quality of service (QoS), severe electromagnetic environments and interference are challenges faced by railway-dedicated mobile communication systems [
2]. Moreover, the non-stationary characteristics of railway channel, which belong to short-term fading behavior, are closely related to dynamic channel modeling, especially for HSR [
3]. All in all, railway radio channel models describe how channel fading behaves in a given scenario, and helps design communication systems and evaluate link- and system-level performance, and the study of channel modeling for railway communication is of great importance.
1.1. Related Works
Scholars have conducted a series of related work with respect to the main railway line channel modeling. Related literature will be reviewed from four aspects.
From the point of view of a communication scenario, different terrains such as railway tunnels, viaducts, cuttings, urban areas, rural areas, as well as different communications scenarios such as train-to-train communication, in-train communication, air-to-ground communication, dynamic channel model with overhead line poles, are all covered by open literature. In [
4], the tunnel entrance scenario channel time correlation function is analyzed based on measurement data. In [
5], based on measurements under different viaduct scenarios at 930 MHz conducted along the “Zhengzhou–Xian” HSR in China, the authors analyzed the impacts of the viaduct height as well as the number of surrounding scatterers on the channel characteristics. In [
6], the authors conducted measurements in the cutting scenario along the “Zhengzhou–Xian” HSR, based on which the scattering components are modeled by clusters.The authors in [
7] focused on the short-term fading behavior of HSR channel with multiple scenarios, such as the rural, station, as well as suburban. In [
8], the train-to-train scenarios are studied, considering typical environments. In [
9], the authors studied the indoor wireless channels in HSTs, and the specific interior layout of the train is considered. In [
10], the air-to-ground channel at 1.4 GHz is discussed, and the altitude of the transmitter is 20 km or 100 km, and the receiver is placed on the train. In [
11], the authors proposed the impact of the overhead line poles—which constitute the infrastructure of the power supply system built along the railway line—on the line-of-sight path of railway channel propagation.
Talking about the frequency bands, the higher the frequency, the greater the path loss, and as a broad consensus, countries around the world generally allocate medium- and low-frequency resources for railway mainlines. As a result, academic research into the frequency characteristics of the railway main lines mostly focus on sub 6 GHz. Ref. [
12] analyzed the performance of the Indonesia Railway Channel Model, with a frequency of 873–880 MHz and 918–925 MHz for up-link and down-link, respectively. In [
13], based on measurements in railway scenarios at a frequency of 930 MHz, the authors presented a general stochastic fading channel model. The authors in [
14] created a new 1.905 GHz channel model for HSTs derived from a measurement campaign conducted on the Beijing–Tianjin commercial HST in China. In [
15], measurements of railway cuttings at 950 MHz and 2150 MHz were carried out, and the broadband measurement results were analyzed. In [
16], measurements along the Harbin–Dalian passenger dedicated railway line of China were conducted at 2.6 GHz with 40 MHz bandwidth, based on which the single-input–single-output (SISO) channel characteristics were derived. In [
17], the authors conducted measurements at 2.7 GHz and 5.6 GHz on a German high-speed track with train speed of 300 km/h, and CIR, the coherence bandwidth, as well as RMS delay are deeply analyzed. Ref. [
18] focused on 3.5 GHz channel characteristics for 5G in urban rail station.
Moving on to research methodology, measure campaign, ray-tracing simulations, machine-learning, deep learning, signal-to-noise ratio quantization strategy, finite-state Markov modeling, multi-task learning, as well as passive channel sounding all have been studied by scholars. In [
19], the wireless channels fading statistics of HSR cutting scenarios are characterized by conducting measurements, and described by a hidden Markov mode. In [
20], the tapped delay line (TDL) models of ray-tracing simulations are used to evaluate the channel state as well as the throughput of 5G wireless communication systems in various HSR scenarios. In [
21], machine learning approaches were used to solve the problem of wireless channel scenarios identification. In [
22], machine learning (ML) was adopted to investigate the MPCs clustering in typical HSR scenarios. In [
23], a deep learning (DL) method was used to analyzed the spatial-temporal prediction of channel state information (CSI) as well as channel statistical characteristics (CSCs) for the future smart HSR communication network. In [
24], the signal-to-noise ratio (SNR) threshold was exploited by authors to established a novel finite-state Markov chain (FSMC) optimization simulation model, by which, the universality and accuracy of channel models can be improved in different HSR scenarios. In [
25], the small-scale fading channels of the “Luoyang South–Lintong East” HSR in China were characterized via the finite-state Markov channel (FSMC) modeling. In [
26], a multi-task learning (MTL) convolutional neural network (CNN) was adopted by the authors to propose a novel super-resolution (SR) model for generating channel characteristics data. In [
27], a passive channel sounder was used to extract 3.45 GHz 5G network’s channel impulse responses (CIRs).
In terms of research object, except channel parameters such as power delay profiles (PDPs), path loss, Rician K-factor, RMS delay spread, re-configurable intelligent surface (RIS), multi-link channels, narrow-beam channel, Doppler shift, as well as propagation scenario identification, all have been involved. In [
28], the channel characteristics of RIS-assisted near-field communication was extensively analyzed, and a 3D RIS-assisted MIMO channel model was proposed. In [
29], a Markov-based multi-link tapped-delay line model for railway multi-link transmission communications is established. In [
30], common and uncommon clusters for the different links of HSR narrow-beam channels are described with a non-stationary 3D wide-band geometry-based stochastic model. Ref. [
31] focused on the influence of mobility of train in rapidly time-varying channels. In [
32], a new propagation scenario identification model was proposed using the long short-term memory (LSTM) neural network.