The rapid development of high-speed railway (HSR) and train-ground communications with high reliability, safety, and capacity promotes the evolution of railway dedicated mobile communication systems from Global System for Mobile Communications-Railway (GSM-R) to Long Term Evolution-Railway (LTE-R). The main challenges for LTE-R network planning are the rapidly time-varying channel and high mobility, because HSR lines consist of a variety of complex terrains, especially the composite scenarios where tunnels, cuttings, and viaducts are connected together within a short distance. Existing researches mainly focus on the path loss and delay spread for the individual HSR scenarios. In this paper, the broadband measurements are performed using a channel sounder at 950 MHz and 2150 MHz in a typical HSR composite scenario. Based on the measurements, the pivotal characteristics are analyzed for path loss exponent, power delay profile, and tap delay line model. Then, the deterministic channel model in which the 3D ray-tracing algorithm is applied in the composite scenario is presented and validated by the measurement data. Based on the ray-tracing simulations, statistical analysis of channel characteristics in delay and Doppler domain is carried out for the HSR composite scenario. The research results can be useful for radio interface design and optimization of LTE-R system.
Over the past few years, the world has witnessed the rapid development of the high-speed railways (HSRs) in China. The GSM-R system, which is specially designed and standardized for communication between train and control centers, could provide voice group call, functional addressing, location dependent addressing, enhanced multilevel priority and preemption, train control safety data transmission, and dispatching non-safety data transmission service. However, the GSM-R still has some limitations to meet a diversity of data transmission including security monitoring and maintenance information in real network operation since it is only available to narrowband communications. Thus, the broadband wireless train-ground communication on HSR plays an important role in train operation control, monitoring, and maintenance data transmission and makes a great number of applications possible [
The researches on broadband and narrowband channels in HSR or subway environments have been conducted over the last few years. The small-scale fading behavior, which is defined as signal variations occurring within very short distances, has been discussed in large numbers of papers for broadband mobile communications. Several wireless channel propagation measurement activities have been performed in different kinds of HSR scenarios, such as the train station [
The deterministic modeling technique based on ray-tracing (RT) simulation is also an important method to model HSR channels [
The main contributions of this paper lie in the following aspects: A broadband wireless channel measurement campaign is conducted in HSR composite scenario by using the channel sounding system at the frequencies of 950 MHz and 2150 MHz. Based on the measurement data, we explicate the phenomenon of the variation trends of path loss, PDP, mean excess delay, and the RMS delay spread in 6 typical areas of the composite scenario. Then, the TDL based channel models are formed, which are helpful for the characterization of the HSR time-varying channel. The HSR composite scenario is reconstructed and the 3D RT simulator is validated by the measurement data. Then, the small-scale fading, PDP, Doppler statistical properties, and the correlation coefficients between delay and Doppler domain are investigated. This is helpful for the development of HSR broadband communication system in the foreseeable future.
The rest of this paper is organized as follows. The measurement scenario, channel sounder configuration, and measurement campaign are provided in Section
The measurement system consists of a channel sounder including transmitter (Tx) and receiver (Rx), oscilloscope, and antennas. The measurement specifications are described in detail below.
A channel sounder serves as a systematic device to measure the behaviors of wireless signals in the specific environment [
Channel sounder parameters.
Parameters | Settings |
---|---|
Transmitter | |
Frequency range | 35 MHz–3000 MHz |
Output power | 35 dBm |
Pulse width | 30 ns, 45 ns, and 60 ns |
Pulse period | 1 us |
Rise and fall time | 3 ns |
Intermediate frequency (IF) bandwidth | 100 MHz |
|
|
Receiver | |
Frequency range | 100 MHz–4550 MHz |
IF | 150 MHz |
IF bandwidth | 100 MHz |
Sensitivity | −100 dBm |
Noise figure | 3 dB |
Video output | 1.2 Vpp |
The output demodulated signal of the receiver is sent to the digital oscilloscope, which allows capturing digitized PDP in real time and saving data on the hard disk. In order to meet the requirements of high sampling rate, Infiniium MSO9104A from Keysight is used, which offers a bandwidth up to 1 GHz and channels with responsive deep memory to ensure superior viewing of signals under test.
The channel sounder is equipped with the L-Com HG72714P-090 panel directional antenna whose polarization is vertical, with
Tx and Rx antenna parameters.
Parameters | Settings |
---|---|
Tx antenna | |
Frequency range | 698–960 MHz, 1710–2700 MHz |
Gain | 13 dBi, 14 dBi |
Vertical beamwidth |
|
Horizontal beamwidth |
|
Polarization | Vertical |
Front to back ratio | ≥23 dB |
Dimensions |
|
|
|
Rx antenna | |
Frequency range | 698–960 MHz, 1710–2700 MHz |
Gain | 4 dBi, 7 dBi |
Vertical beamwidth |
|
Horizontal beamwidth |
|
Polarization | Vertical |
Lengths | 1.02 m |
Antenna pattern.
Tx antenna pattern
Rx antenna pattern
In the construction of railway in mountainous region, the cuttings are often used to help the HSR cross the large obstacles. A standard deep cutting with the slopes of both sides having the similar inclination can produce abundant reflection and scattering components. Viaducts with a height of 10 m to 30 m are also used in HSR to ensure flatness and straightness of rails. The height of the antenna on the roof of the train is increased due to viaduct, and this reduces the number of scatterers and forms a relatively “clear” line-of-sight (LoS) propagation channel. Therefore, cutting and viaduct have a pronounced effect on channel propagation characteristics, which provides the motivation for investigating the composite scenario of cutting and viaduct in this paper.
The broadband measurements were performed in Xinzhou section of the “Datong-Xi’an” HSR in China, as shown in Figure
The measured composite scenario in Datong-Xi’an HSR.
Scenario from Google Earth
Field measurement scenario
This paper mainly analyzes and discusses the effects of multipath components in composite scenarios. In order to achieve this goal, a series of broadband measurements at 950 MHz and 2150 MHz bands which may be used for LTE-R in future were conducted. The channel sounder transmitter was installed on the platform at the top of the cutting, the Tx antenna was 35 m high from the rail surface and 20 m away from the tunnel portal. The pitch angle of transmitting antenna was about
The measurement environment can be classified as four regions in the composite scenario: locations 1 and 2 belong to region I, that is, the tunnel portal; locations 3–15 fall into region II which is cutting 1; locations 16–44 belong to region III, where a viaduct exists, and region IV, which is cutting 2, includes the rest of the locations. The sampling interval is 50 ns and approximately 20000 CIRs for each snapshot are collected. Each neighboring snapshot has a distance of 10 m.
The knowledge of wireless channel fading behavior is helpful for designing fading countermeasures. Thus, parameterizing the fading properties and developing channel model for HSR are necessary. This section will make a detailed analysis of the measurement results.
The path loss model assumes a linear dependency between the path loss in dB and the logarithm of distance
Path loss at 950 MHz and 2150 MHz in the composite scenario.
The PDP is widely used for modeling multipath fading channel. The extent of multipath effects on radio channel can be clearly determined by the time delay spread. Channel parameters such as the first and second significant multipath components can be extracted from the PDP curve.
Since the HSR base station is always installed on one side of the track and the antenna is higher than the track by at least 25 m except for tunnel part, railway environment is a typical scenario with LoS propagation. Usually, the radio wave coming from LoS path offers the dominant component, and the radio waves coming from other directions are scattered or reflected by the steep walls or hills, causing the time delay spread.
From Figures
PDP at 950 MHz in the composite scenario.
PDP at 2150 MHz in the composite scenario.
The typical PDPs on different scenarios are demonstrated in Figures
Typical PDP for composite HSR scenario.
Tunnel portal
Cutting 1
Viaduct near cutting 1
Viaduct
Viaduct near cutting 2
Cutting 2
As a general method for analyzing broadband channel measurements, the PDP can be quantified by the mean excess delay and RMS delay spread, which describe the time dispersion characteristics of the channel in a specific scenario. Delay of the multipath component is estimated relatively to the first detectable peak path [
In general, the mean excess delay and the RMS delay spread rely on the selection of noise threshold, which is used to distinguish the received multipath components power from the thermal noise power for raw PDPs. In this paper, the noise threshold is set to be 6 dB above the noise level of PDP, as given in [
In Table
Parameters of time dispersion at two frequencies.
Region index | I | II | III | IV | |
---|---|---|---|---|---|
Corresponding scenario | Tunnel | Cutting | Viaduct | Cutting | |
950 MHz | Mean excess delay (ns) | 374.66 | 193.11 | 60.61 | 31.40 |
950 MHz | Average RMS delay spread (ns) | 218.67 | 125.40 | 50.01 | 40.20 |
950 MHz | Average number of multipaths | 6 | 5 | 1 |
1 |
2150 MHz | Mean excess delay (ns) | 270.68 | 195.68 | 66.31 | 43.62 |
2150 MHz | Average RMS delay spread (ns) | 120.44 | 135.18 | 52.03 | 38.81 |
2150 MHz | Average number of multipaths | 4 | 5 | 1 |
1 |
RMS delay spread at two frequencies.
The TDL based channel model is helpful for analyzing the time-varying channel [
Table
TDL channel model for tunnel portal.
Scenario | Tap | Relative delay (ns) | Average gain (dB) |
---|---|---|---|
950 MHz | |||
Tunnel portal | 1 | 0 | 0.0 |
2 | 119.7 | −29.48 | |
3 | 353.9 | −31.84 | |
4 | 452.0 | −27.25 | |
5 | 548.8 | −25.14 | |
6 | 695.8 | −31.28 | |
|
|||
2150 MHz | |||
Tunnel portal | 1 | 0 | 0.0 |
2 | 77.7 | −14.21 | |
3 | 174.1 | −18.23 | |
4 | 553.5 | −19.5 |
Table
TDL channel model for cutting.
Scenario | Tap | Relative delay (ns) | Average gain (dB) |
---|---|---|---|
950 MHz | |||
Cutting 1 | 1 | 0 | 0.0 |
2 | 82.9 | −36.86 | |
3 | 138.1 | −51.26 | |
4 | 192.6 | −52.93 | |
5 | 241.4 | −55.66 | |
|
|||
2150 MHz | |||
Cutting 1 | 1 | 0 | 0.0 |
2 | 95.7 | −32.16 | |
3 | 180.4 | −55.39 | |
4 | 254.7 | −57.35 | |
5 | 305.1 | −55.71 |
Table
TDL channel model for viaduct of both ends’ connecting cuttings.
Scenario | Tap | Relative delay (ns) | Average gain (dB) |
---|---|---|---|
950 MHz | |||
Viaduct near cutting 1 | 1 | 0 | 0.0 |
2 | 108.3 | −37.27 | |
|
|||
Viaduct near cutting 2 | 1 | 0 | 0.0 |
2 | 89.5 | −9.5 | |
|
|||
2150 MHz | |||
Viaduct near cutting 1 | 1 | 0 | 0.0 |
2 | 106.5 | −22.95 | |
3 | 428.4 | −39.6 | |
|
|||
Viaduct near cutting 2 | 1 | 0 | 0.0 |
2 | 107.3 | −21.46 |
As can be seen from Figure
As a deterministic model, the propagation is greatly affected by the modeling of the scenario. As a result, precisely reconstructing the 3D propagation environment is the key factor for deterministic channel modeling. With the help of 3D modeling, the radio channel behavior of the composite scenario has been simulated by the deterministic propagation model. To begin with, the dimensions of the viaduct, cutting, and buildings are measured manually. Then, the latitude and longitude information of the test line is collected and the scenario around the test line is obtained by the satellite image of Google Earth. Finally, by gathering all the information, a complete 3D RT channel model library can be established.
In light of different radio propagation mechanisms, all the major objects in HSR scenario could be divided into the large-scale and the small-scale structures. The large-scale structures, including ballastless track, tunnel, cutting, and viaduct, are modeled as triangles and polygons with different electromagnetic parameters. The small-scale structures, including pylons and vegetation, are modeled as a conductive cylinder with a limited length. The scenario is composed of tunnel, mountain, cutting, viaduct, and ballastless track. It is essential to know the dielectric properties of materials for improving the accuracy of simulation. The reflection coefficients and scattering coefficients of all the materials above can be determined by the following approaches. First of all, the standardized parameters are defined in ITU-R recommendation, like ITU-R P.2040, ITU-R P.1411, ITU-R P.1238-7, and so on. These parameters can be set to the original inputs. Furthermore, for specific environments where materials are unavailable or insufficient in standards, measurements will be used to calibrate the coefficients. In this work, we apply the proposed approach in [
Reconstruction of the composite HSR scenario.
The 3D ray-optical based channel simulation is used to establish 3D deterministic channel model. Several propagation mechanisms are taken into account to model the multipath nature of HSR channel. Firstly, the direct path between the Tx and Rx is calculated based on the distance and path loss. Secondly, specular reflection and diffuse scattering are calculated, with the bouncing order of reflection up to 2 for reducing the computational complexity. Furthermore, the contribution of diffraction is neglected because there is no obstacle between the Tx and Rx. Thus, the information of each ray can be obtained, including the type of path, path loss, path delay, and 3D angular properties. The positions of Tx/Rx, the directions of the antennas, and a snapshot of the RT result are illustrated in Figure
A snapshot of the RT simulation.
For the purpose of improving RT algorithm efficiency, the objects involved in the deterministic channel model are dominant for the HSR, such as the viaduct, tunnel, and cutting. Hence, the model has excluded the objects that are remote, unobservable, and lower than the track. In RT algorithm,
Figure
Comparison of path losses between the measured and the simulated results.
Contrast of the PDP based on the measurement and the deterministic modeling.
950 MHz
2150 MHz
Based on the above 3D scenario, small-scale fading characteristics are investigated through the channel simulation in this subsection. The signal bandwidth, power, antenna parameters, and location are the same as field measurement. Figure
Analysis of normalized small-scale fading.
Zone index | 1 | 2 | 3 | 4 | 5 |
Time segment in s | 0 |
1.5 |
2.7 |
4.7 |
5.6 |
Corresponding scenario | Tunnel portal and cutting | Viaduct | Viaduct | Joint of viaduct and cutting | Cutting |
Fading depth in dB (950 MHz) | 6.1 | 5.2 | 2.1 | 0.7 | 0.9 |
Maximum fading depth in dB (950 MHz) | 46.07 | 14.87 | 6.02 | 1.68 | 4.38 |
Fading depth in dB (2150 MHz) | 15.4 | 17.9 | 2.8 | 15.5 | 0.7 |
Maximum fading depth in dB (2150 MHz) | 48.37 | 37.87 | 3.35 | 59.04 | 1.37 |
Small-scale fading.
950 MHz
2150 MHz
Wireless communication system must be developed for various propagation conditions, which motivates stochastic characteristic analysis that treats a linear time-varying channel as a random quantity. Since the channel has no great change in the stationarity interval which is assumed to be 10 ms [
Based on the RT simulation, the PDP is studied in this subsection. Figure
PDP versus scenario time at different frequencies.
950 MHz
2150 MHz
Based on the RT simulation, Doppler statistical properties are analyzed in this subsection. The mean Doppler shift
Doppler power spectrum versus scenario time at different frequencies.
950 MHz
2150 MHz
RMS Doppler spread and mean Doppler shift versus scenario time at different frequencies.
950 MHz
2150 MHz
Based on the RT simulation, the cross-correlation coefficient between delay and Doppler domain is estimated in this subsection, which is defined as [
The correlation coefficients between delay and Doppler domain.
Scenarios | Correlation coefficients |
---|---|
950 MHz | |
Tunnel portal and cutting 1 | 0.5469 |
Viaduct | 0.0308 |
Cutting 2 | 0.1011 |
|
|
2150 MHz | |
Tunnel portal and cutting 1 | 0.5082 |
Viaduct | 0.0143 |
Cutting 2 | 0.3032 |
In this paper, the broadband wireless channel measurements using a customized channel sounder in a real HSR composite scenario at 950 MHz and 2150 MHz are reported and analyzed. The path loss exponent and parameters of time dispersion containing the number of paths, the mean excess delay, and the RMS delay spread are estimated. The TDL channel model is proposed for the HSR composite scenarios containing tunnel portal, cutting, and viaduct based on the actual measurement data. The analysis reveals some important phenomena for the composite scenario: the difference of RMS delay at two frequencies is very small while the propagation loss is more serious; the cutting and tunnel portal will bring greater RMS delay; the multipath in viaduct is more influenced by the two cuttings that are connected to the viaduct, because the interference from the reflection and scattering of the cuttings produces not only deep fading, but also Doppler dispersion and delay spread. The deterministic channel model of the composite scenario, which is reconstructed using the 3D RT method, is put forward in this paper. The channel model is compared and validated with measurement data. The transition regions of different scenarios could be clearly identified by the normalized small-scale fading and Doppler characteristics analysis. Our future work will continue to analyze cluster birth and death behavior of the multipath components in this scenario. As radio channels have profound impacts on the field strength coverage, reliability, and quality of service of HSR mobile communication system, the results are useful for providing guidance on the broadband HSR communication system design, network planning, and optimization.
The authors declare that there are no conflicts of interest regarding the publication of this paper.
The authors express their thanks to the support in part from the National Key R&D Program of China under Grant no. 2016YFB1200102-04, the National Natural Science Foundation of China under Grants 61501021, U1334202, 61501020, and 61771037, the Project of China Railway Corporation under Grants 2016X009-E, 2016X003-L, and 2016X003-O, the Fundamental Research Funds for the Central Universities under Grant 2016JBM076, State Key Lab of Rail Traffic Control and Safety Project under Grant RCS2016ZJ005, and Beijing Natural Science Foundation under Grant L161009.