Keywords

1 Introduction

In high-speed railways (HSR) the throughput of the wireless systems is limited by speed of the vehicle. The current world speed record for HSR is 574.8 km/h and was achieved by an Alstom train on a SNCF track in 2007. As we move forward to 5G, the research on wireless faces new challenges. The most critical one is how to provide a high capacity wireless system on a HSR scenario, to support railway services like ERTMS, real-time CCTV, infotainment, Internet access for passengers and a large etcetera [1].

One of the most important features of HSR scenarios is the fast fading caused by the large relative speed (this is, Doppler spread) [2]. Moreover, Doppler spreads are time-varying so it means also a non-stationary fading channel. The Doppler shift causes a misalignment of the frequencies of transmitter and receiver. This misalignment can decrease the subcarrier orthogonality and cause Inter-Carrier Interference in OFDM systems. Moreover, as the speed of the train increases, the assumption of having a perfect knowledge of the channel (as in low mobility scenarios) is no longer valid [2]. This means that the performance of the whole wireless system could decrease significantly. For all these reasons, channel modeling techniques are very important in an HSR environment.

Several measurement campaigns have been taken in different HSR scenarios, like viaducts [3], hilly terrains [4] cuttings [5], train stations [6] and tunnels [7]. The empirical results analysis and statistical modeling in large-scale and small-scale have been conducted based on these measurements.

The structure of this article is as follows: Sect. 2 provides an overview of the HSR environment where the measurements were taken; Sect. 3 presents the results and in Sect. 4 the conclusions are drawn.

2 Environment

Most of the research work aforementioned is dedicated to explain a unique scenario (cutting, or viaduct, or tunnel, etc.). In fact, real railway lines are formed by the combination of all these unique scenarios. This is of special importance when the train moves at a high speed, because the transition from one scenario to other is done very quickly. So, these regions with different topographies cannot be separated easily. We say that a scenario formed by many unique scenarios is a composite scenario. The existing research has not considered the transitions between two scenarios. Therefore, the non-stationary of fast varying channels in these HSR composite scenarios requires further research.

To measure the wireless channel a portable channel sounder is used. It operates generating a train of periodic narrow pulses. The distorted signal is received by the channel sounder receiver. After the envelope detection, the demodulated signal is sent to the digital oscilloscope (Keysight Infiniium MSO9104A). Further details about the channel sounder can be found in Table 1.

Table 1. Channel sounder parameters
Full size table

Moreover, the channel sounder transmitter is equipped with a directional antenna (L-Com HG727 14P-090); whereas on the receiver side we have an omnidirectional antenna (L-Com HG72107U). Both of them are vertical polarized (see Table 2 for further details about the antennas).

Table 2. Antenna parameters
Full size table

The composite scenario where the measurements were taken has a viaduct and a deep cutting. In Fig. 1 a sketch of the composite scenario is depicted. The cuttings shape can imply a significant number of reflections and some scattering as well. Viaducts (usually with a height of 10–30 m) are also used in HSR to avoid slopes in the track which undermine train’s high speed. It is very common to raise the antennas in the viaduct to maintain the LOS between transmitter and receiver, as well as decreasing the number of scatters that can affect the signal. Therefore, cutting and viaduct have opposite effects on the channel. All these effects imply the impossibility to model the propagation in a HSR environment as a pure LOS channel.

Fig. 1.
figure 1

Composite scenario. Tunnel entrance, cutting, viaduct and another cutting.

Full size image

The wideband measurements have been conducted in a composite scenario, which is located in the “Datong-Xi’an” HSR line (Xinzhou, Shanxi Province, China). The composite scenario starts at the end of a tunnel and is formed by a 130-m long cutting (immediately after the tunnel), followed by a viaduct of 250 m and, finally another cutting of 100 m (Fig. 1, from right to the left). So we have four regions in our scenario: region 1, near to the tunnel, where the cutting causes a lot of multipath; region 2, the first cutting where the two steep walls can reflect the transmitting waves and create some multipath; region 3, the viaduct, with almost no multipath (is an open area); and finally, region IV where the second cutting is located.

This composite scenario is very representative of the typical HSR environments where the main scatters are located near the track. The receiver is located on the track 30 m away from the transmitter (see Fig. 2). To take the measurements the receiver is moved along the track taking average power-delay-profiles at 50 different positions (distance between transmitter and receiver varies from 30 m to 458 m). To evaluate the effects of multipath on the channel in a composite scenario like this one we have taken a set of wideband measurements at 950 MHz and 2150 MHz.

Fig. 2.
figure 2

Measurement setup. The receiver is over the track and the transmitter is 30 m away (on the right)

Full size image

3 Results

We provide two different results: power-delay profiles and time dispersion analysis at the two frequencies considered (950 and 2150 MHz). Power-delay profiles are very useful to characterize the fading.

3.1 Power-Delay Profile

In the area close to the tunnel a NLOS scenario is presented in region 1. PDPs at both frequencies display a weak main component and a large delay (see Fig. 3a). In particular, the delay spread at 950 MHz is larger and more multipath happens at 950 MHz than at 2150 MHz. In region 2 we are in a stronger LOS scenario with little delay spread (see Fig. 3b). In region 3 the LOS component is attenuated, but the impact of cutting 1 is still present. Therefore, the PDP in Fig. 3c has similar multipath components with also weak dominant LOS components. In Fig. 3d, which corresponds also to region 3 (viaduct) multipath components are very weak due to the lack of surface to reflect the signal, and consequently, the delay spread is much lower. Once we move to the other side of the viaduct towards cutting 2, the LOS components decrease as the separation between transmitter and receiver increases. Equally, we start receiving reflections from cutting 2, which implies more delay spread (see Fig. 3e). In cutting 2 there is only one path (see Fig. 3f). This is because the size of this cutting is smaller than cutting 1 and has an asymmetry with lower height, so the time delay caused by the multipath components is very short and close to the LOS component.

Fig. 3.
figure 3

Power-delay profiles in (a) tunnel entrance; (b) cutting 1; (c) viaduct 1 (the part of viaduct close to cutting 1); (d) middle of the viaduct (viaduct 2); (e) the part of viaduct close to cutting 2 (Viaduct 3); (f) cutting 2.

Full size image

3.2 Time-Dispersion Analysis

To analyze the delay spread on a wideband measurement, a valid method is to estimate the excess delay and RMS delay spread. These two parameters characterize the time dispersive nature of the channel and can as well be used to determine the number of channel taps. The excess delay is the maximum delay above a threshold which reveals the maximum relative distance between the receiver and the reflection or scattering surfaces. The RMS delay spread is the square root of the second central moment of the power-delay profile.

The average excess delay, average RMS delay spread and average number of channel taps in each region are summarized in Table 3. The RMS delay along the measurement points at 950 MHz and 2150 MHz is presented in Fig. 4.

Table 3. Time dispersion results
Full size table
Fig. 4.
figure 4

RMS delay spread at 900 MHz and 2150 MHz

Full size image

The variation of the RMS delay spread through the track is interpreted as an evidence of the aforementioned results. The multipath components on each region of the composite scenario are different. In particular as we move from the entrance of the tunnel and cutting 1 (region 1) to an open area (viaduct), the average RMS delay decreases 1 magnitude order (from 70–100 ns to 5–7 ns). Also the number of channel taps decreases significantly. However, average excess delay in region 4 (cutting 2) is smaller than in region 3, but the RMS delay values are just the opposite.

4 Conclusion

We present some measurements on a composite scenario of a high-speed railway line. This composite scenario is very representative as this type of lines are not straight and flat countryside tracks, but more complex layouts with viaducts, tunnels, cuttings and other geographical features. The most relevant parameters to analyze the time dispersion properties have been extracted. These parameters are the excess delay, the RMS delay spread, and the number of channel taps. Furthermore, the TDL channel models can be established in detail in different regions.