Railway transportation, as an important pillar of national economics, is directly related to the safety of people’s lives and property and is hailed as the “artery of the country”. As a common consensus, safety is the lifeline of railway transportation. Railway radio communication plays a very important role in railway transportation with respect to transportation production as well as safety management; as a means of communication and information technology, railway radio communication reliably carries the dispatching communication service, train control information, etc., and it is the foundation of promoting production efficiency and the high-quality development of railways.
Radio frequency is the foundation and prerequisite for the innovation and development of railway radio-communication technology. In view of the high reliability requirements of railway radio-communication systems, it is widely recommended by the International Union of Railways (UIC) and other countries in the world to allocate dedicated frequency resources for railway radio-communication systems. The International Telecommunication Union (ITU), UIC, and other countries have paid close attention to the coordination of railway-dedicated frequencies and frequency protection. In 2015, the World Radio-communication Conference (WRC-15) decided to set up agenda item 1.11 to study spectrum harmonization for railway radio-communication systems between train and trackside within the existing mobile service allocations, and countries conducted in-depth research on railway radio-communication systems at the ITU level. In UIC, under the umbrella of the future railway mobile communication system (FRMCS) project, the Frequency Working Group (UGFA) was established and is now responsible for studying the related topics of frequency with respect to railway radio-communication systems.
The dedicated frequency for railway radio communication has played a significant role in railway transportation production, ensuring safety and improving efficiency, as well as achieving huge economic and social benefits. The European Union has allocated 876–880 MHz (train to ground) and 921–925 MHz (ground to train) as dedicated operating frequency bands for GSM-R in Europe, with over 100,000 km of lines currently deployed. In China, GSM-R uses 885–889 MHz (train to ground) and 930–934 MHz (train to ground), and covers over 85,000 km of railway lines.
With the trend of digitalization and intelligence, the contradiction between the increasing service demand and the shortage of frequency resources as well as the limited carrying capacity of railway radio-communication systems has become increasingly prominent. Facing the rapid evolution of radio-communication technology, the broadband of railway radio-communication systems has become a future development direction. Considering the limited frequency resources in the low-frequency band and fragmented allocation, it is an inevitable trend to migrate to the middle- and high-frequency band. In Europe, the existing 2 × 4 MHz R-GSM frequency resources of GSM-R are expanded to the 2 × 5.6 MHz (874.4–880/919.4–925 MHz) ER-GSM frequency band; in addition, 1900–1910 MHz is used as a supplementary frequency band for FRMCS(5G-R). In the meantime, China Railway has submitted an application to the Ministry of Industry and Information Technology(MIIT) for using the 2100 MHz 2 × 10 MHz frequency band as a 5G-R system field test frequency.
However, in addition to meeting the requirements of services on the main lines, such as dispatching communication, train control, monitoring, etc., service models in hotspots such as stations and yards are more complicated in terms of service types as well as bandwidth requirements. The 5G-R system facilitated with dedicated frequency support cannot meet the entire communication requirements of hotspots, so other radio-communication technologies and methods need to be adopted as a supplement to the 5G-R communication system. Among these, the mmWave communication system is a promising technology, especially for large bandwidth communication between train and trackside.
1.1. Related Works
Scholars have carried out a series of related work with respect to mmWave. Corresponding research will be reviewed from four categories involving communication scenarios, frequency bands, and research methodology, as well as the object of the research.
From the communication scenario point of view, railway, metro, residential house, indoor commercial, tropical area, and many other scenarios are covered by the open literature. With respect to railway scenarios, in [
3], channel characteristics are studied in the 5G mmWave band centered at 25.25 GHz for typical urban HSR scenarios, including straight and curved route shapes. In [
4], the 28 GHz mmWave channel characteristics of rural railway scenarios are studied via a calibrated ray-tracing simulator. Ref. [
5] also focuses on the 28 GHz mmWave channel characteristics of rural railway scenarios, in which the dominant propagation mechanisms (direct, penetration, reflection, scattering, etc.) are determined and the three-dimensional environment model and the electromagnetic parameters of different objects are calibrated. The automatic coupling application scenario is studied in [
6], in which a dynamic train-to-train (T2T) mmWave propagation measurement campaign was conducted, and the received signal power in the open field area was analyzed and modeled using a two-ray path loss model. The results indicated that the received power next to a platform was higher compared to the open field counterpart due to strong contributions from the signals reflected by the platform. The work in [
7] studied the wireless coverage for intra-wagon scenarios at the 60 GHz band; by considering the balance of the wireless coverage and the cost of the transmitters, the deployment of two transmitters in one wagon is suggested. Typical objects in the railway propagation environment are also important research areas. In [
8], a significant analysis methodology, which analyzes and distinguishes the contributions of different railway objects to the mmWave propagation channel, is proposed. In [
9], the influence of typical objects to the mmWave propagation channel is analyzed for “train-to-infrastructure” and “intra-wagon” railway scenarios with various configurations. In [
10], a statistical mmWave channel modeling for railway communications backhaul in 5G networks was proposed, involving 28 GHz and 60 GHz, by simulating in a novel software, NYUSIM. Viaducts and tunnels are common scenarios of railway terrain, especially in high-speed railways. The work in [
11] considered measurement-based ray-tracer calibration and channel analysis for high-speed railway viaduct scenarios at 93.2 GHz, and the conclusion that the typical structure of viaducts and the application of horn antenna lead to a small value of Rician K-Factor and RMS delay spread is drawn. In [
12], the authors investigate the mmWave propagation characteristics of a high-speed moving train based on field measurements in tunnel and viaduct scenarios. The measurements were carried out at 28 GHza and path loss (PL) and other channel parameters, including delay spread and Doppler shift, were investigated. In [
13], smart-rail mobility was discussed, and the authors identified the main technical challenges and the corresponding chances concerning the reference scenario modules, accurate and efficient simulation platforms, beamforming strategies, and handover design. A mmWave beamforming scheme for disaster detection in high-speed railways was proposed in [
14]; the key point of the study was that the antenna array generates multiple beams with different beam widths in different frequency bands simultaneously and the multiple beams are responsible for different detection areas. Another way to improve the reliability of railway communication is interlaced redundant coverage; in [
15], a location-fair-based mmWave stable beamforming scheme under interlaced redundant coverage architecture is proposed to improve the stability and reliability of high-speed railway communications. Hotspots, such as depots and shunting yards, are also important railway scenarios. The work in [
16] presented a comprehensive study on millimeter-wave-based mobile hotspot network (MHN) systems for high-speed train communications, including system design, field trial, channel modeling based on measurement campaign, simulation, and validation. In [
17], mmWave channel measurements are reported for a railway depot environment using a wideband channel sounder operating at 60 GHz; path loss, RMS delay, as well as K-factors were extracted. The vacuum tube ultra-high-speed train (UHST) is a hot topic worldwide, and is a potential development direction of future transportation. In [
18], a three-dimensional non-stationary mmWave geometry-based stochastic model (GBSM) is proposed to investigate the channel characteristics of UHST channels in vacuum tube scenarios.
Concerning the frequency bands, 26 GHz, 28 GHz, 30 GHz, 38 GHz, 40 GHz, 60 GHz, 73 GHz, 90 GHz, 93.2 GHz, and up to 300 GHz have all been studied by scholars, among which, 28 GHz and 60 GHz seem to have a higher level of attention. In [
19], the authors conducted a 28 GHz-band 5G experimental trial on the actual Shinkansen that runs at a maximum speed of 283 km/h, and the experimental trial could achieve throughput exceeding 1.0 Gbps and consecutive handovers among the three BSs. In [
20], field experiments of a 28 GHz-band 5G system at indoor train station platforms were carried out, and through the actual measurements, the results confirmed that the prototype 5G system could achieve mobile broadband capacity (more than 1 Gbps), even when the UE was located anywhere at the indoor train station platform. Ref [
4] focuses on channel characteristics in rural railway environments at 28 GHz, and the mmWave channel characteristics of rural railway scenarios are studied via a calibrated ray-tracing simulator. In [
5], influence analysis of typical objects in rural railway environments at 28 GHz is performed. In [
21], millimeter-wave (mmWave) high-speed train measurements were conducted at 28 GHz with a speed up to 170 km/h in two different HST scenarios, viaduct and tunnel, in which large- and small-scale fading characteristics were extracted.
Moving on to research methodology, measure campaigns using measurement results published in the open literature, ray-tracing simulations, ultrawideband (UWB) channel sounding, deep learning, deep reinforcement learning (DRL), machine learning(ML), the space alternating generalized expectation-maximization (SAGE) algorithm, simulation platforms such as MATLAB, NYUSIM, mathematical methods such as non-convex problem with a near-optimal solution, coalition game, as well as the Bayesian approach, have all been used by scholars. Among all these methodologies, measure campaigns are basically very costly and time-intensive [
22]; thus, ray-tracing simulations calibrated with measured data are commonly used, which means only a relatively small number of channel samples can be obtained, and then measurements are thus used to quantify the accuracy of a ray tracer. In [
9], propagation measurements are conducted in the mmWave band for the 12 most common railway materials, the corresponding electromagnetic parameters are obtained, and a 3D ray-tracing (RT) simulator is calibrated. In [
23], a series of horizontal directional scan-sounding measurements are performed inside a real high-speed train wagon at 60 GHz and 300 GHz frequency bands, and the channel characteristics are extracted, based on which, a self-developed RT simulator is validated through the reconstruction of the three-dimensional wagon model and the calibration of the electromagnetic properties of the main materials. In [
5], based on the channel measurement conducted at 28 GHz with train-to-infrastructure deployment in the rural railway environment, a ray tracing (RT) simulator is calibrated in terms of environment modeling and electromagnetic calculation. In [
11], ray tracer (RT) is employed to simulate the propagation in a high-speed railway viaduct scenario at a frequency of 93.2 GHz. In the article, by comparing the path loss between simulation and measurement, the permittivity and loss tangent of concrete, which is the construction material of viaducts, are calibrated in the ray tracer. In [
24], based on the wideband measurements conducted in the tunnel scenario by using the “mobile hotspot network” system, 3D ray tracing (RT) is calibrated and validated to explore more channel characteristics in different HSR scenarios. In [
25], since the measurement of a mmWave channel is deficient for high mobility, the proposed channel model is developed using a ray-tracing (RT) simulator that is validated with the channel measurements performed in the HST scenario at 93.2 GHz. Ref. [
26] focuses on the analysis of propagation characteristics for train–ground communication systems in tunnel scenarios at both low-frequency and mmWave bands, based on ray-tracing (RT) simulation, and the material parameters in the RT simulation are calibrated by measurement data collected in realistic tunnel environments.
As for the research object, except for channel parameters such as power delay profiles (PDP), path loss, Rician K-factor, root-mean-square (RMS) delay spread, azimuth spread of arrival, and azimuth spread of departure, reconfigurable intelligent surface (RIS), beamforming, Doppler shift, energy efficient, user association, resource allocation and computation offloading, the influence of meteorological attenuation, sensing, minimal distance between base stations, as well as mmWave network architectures for railway have all been involved. In [
27], the authors proposed a RIS-assisted scheduling scheme for scheduling interrupt flows and improving quality of service (QoS), and in the proposed scheme, an RIS is deployed between the BS and multiple mobile relays (MRs). In [
28], multiple beams with different beamwidths are formed by the base station (BS) simultaneously to improve the system capacity, and the mobile relays (MRs) are provided with the ability to adjust the receiver (RX) beams automatically to enhance the received signal-to-noise ratio (SNR). In [
29], control information and user information are transmitted through the high-frequency micro base station (BS) and the low-frequency macro BS, respectively, and an optimized beam width and power allocation scheme is proposed, which is combined with mobile relay (MR) technology to utilize the architecture of large-scale antenna beamforming. In [
30], dynamic and fixed beamforming is evaluated based on the generalized model and the measured data, and the results show that the average throughput of dynamic beamforming is only 4% higher than that of fixed beamforming in the HSR tunnel, but 21% higher in the train station when severe beam misalignment is present. In [
31], the authors proposed a new Doppler shift estimator for mmWave communication systems on HSRs: an equally-divided structure-based estimator (ESBE) that divides the effective orthogonal frequency-division multiplexing (OFDM) symbol into multiple equal fragments. In [
32], the modeling of the Doppler effect for mmWave in HSR communications is conducted, and data-aided Doppler estimation and compensation algorithms are designed based on the new model. In [
33], the authors proposed a new machine learning-based Doppler shift estimator (MLDSE), which estimates the Doppler shift by using the reference signal received power (RSRP) values measured by the mobile receiver at all times.