Portal de Periódicos da CAPES

Path loss models for train‐to‐train communications in typical high speed railway environments

2018; Institution of Engineering and Technology; Volume: 12; Issue: 4 Linguagem: Inglês

10.1049/iet-map.2017.0600

1751-8733

Paul Unterhuber, Stephan Sand, Uwe‐Carsten Fiebig, Benjamin Siebler,

Millimeter-Wave Propagation and Modeling

IET Microwaves, Antennas & PropagationVolume 12, Issue 4 p. 492-500 Special Issue: Selected Papers from the 11th European Conference on Antennas and Propagation (EuCAP 2017)Free Access Path loss models for train-to-train communications in typical high speed railway environments Paul Unterhuber, Corresponding Author Paul Unterhuber paul.unterhuber@dlr.de Institute of Communications and Navigation, German Aerospace Center (DLR), Oberpfaffenhofen, 82234 Wessling, GermanySearch for more papers by this authorStephan Sand, Stephan Sand Institute of Communications and Navigation, German Aerospace Center (DLR), Oberpfaffenhofen, 82234 Wessling, GermanySearch for more papers by this authorUwe-Carsten Fiebig, Uwe-Carsten Fiebig Institute of Communications and Navigation, German Aerospace Center (DLR), Oberpfaffenhofen, 82234 Wessling, GermanySearch for more papers by this authorBenjamin Siebler, Benjamin Siebler Institute of Communications and Navigation, German Aerospace Center (DLR), Oberpfaffenhofen, 82234 Wessling, GermanySearch for more papers by this author Paul Unterhuber, Corresponding Author Paul Unterhuber paul.unterhuber@dlr.de Institute of Communications and Navigation, German Aerospace Center (DLR), Oberpfaffenhofen, 82234 Wessling, GermanySearch for more papers by this authorStephan Sand, Stephan Sand Institute of Communications and Navigation, German Aerospace Center (DLR), Oberpfaffenhofen, 82234 Wessling, GermanySearch for more papers by this authorUwe-Carsten Fiebig, Uwe-Carsten Fiebig Institute of Communications and Navigation, German Aerospace Center (DLR), Oberpfaffenhofen, 82234 Wessling, GermanySearch for more papers by this authorBenjamin Siebler, Benjamin Siebler Institute of Communications and Navigation, German Aerospace Center (DLR), Oberpfaffenhofen, 82234 Wessling, GermanySearch for more papers by this author First published: 27 February 2018 https://doi.org/10.1049/iet-map.2017.0600Citations: 17AboutSectionsPDF ToolsRequest permissionExport citationAdd to favoritesTrack citation ShareShare Give accessShare full text accessShare full-text accessPlease review our Terms and Conditions of Use and check box below to share full-text version of article.I have read and accept the Wiley Online Library Terms and Conditions of UseShareable LinkUse the link below to share a full-text version of this article with your friends and colleagues. Learn more.Copy URL Share a linkShare onFacebookTwitterLinkedInRedditWechat Abstract Within the next few decades, railways will undergo a major modernisation in order to increase their efficiency and safety. The modernisation process will enable new concepts such as fully autonomous trains and dynamic coupling of consists. Today's centralised railway management systems will not be able to safely handle these train concepts and, thus, will be replaced or complemented by new systems. These systems require a tight train monitoring which is based on the reliable exchange of information covering among others the accurate position and velocity of trains. The authors are sure that modern train concepts demand both reliable train-to-train (T2T) communication and improved capabilities for train-to-ground communication. However, T2T communication is hardly investigated and so far no technology has been selected. With this publication, the authors like to provide more insight into T2T signal propagation. This is the basis for the future development of a modern T2T communication system. The authors focus on wide band propagation for T2T scenarios and describe an extensive channel sounding measurement campaign involving two high speed trains. Results on signal propagation measurements in high speed T2T scenarios are presented for different environments. Hence, path loss models for rural, sub-urban and tunnel environments are introduced. 1 Introduction Fully autonomous vehicles will be the future on our streets as well as on the railway tracks. Several manufacturers are testing prototypes on the road and national railway operators are promising driverless train operation within the next 5 years. Hence, the interest in vehicular communications has been increasing tremendously in the last few years. The challenging propagation environment for vehicle-to-vehicle (V2V) communication in road and railway traffic demands comprehensive investigations of the propagation channel and the development of new channel models. Signal propagation and communications for car control and monitoring are well investigated research topics; a first communication standard referred to as intelligent transport systems (ITS-G5) has been defined for V2V communications. Signal propagation and communication for railway control and monitoring are also widely acclaimed research topics. Nevertheless, the focus is almost exclusively on train-to-ground (T2G) systems. A major achievement has been the T2G communication standard global system for mobile communications-railway (GSM-R); long term evolution has been investigated as a possible successor of GSM-R [[1]]. Train-to-train (T2T) signal propagation and communication are topics which are hardly touched by any research agenda in the past. As a consequence, the propagation aspects of T2T links are not investigated except for low frequency and medium speed [[2]]. A survey of channel measurements and models on intra-train, intra-consist and T2T signal propagation has been presented in [[3]]. New railway concepts like virtual coupling can increase efficiency and reliability of service. In [[4]] virtual coupling is described in detail as a technique, where position and speed information is transmitted between trains with very low latency. It allows trains to drive on the same track with a shorter quasi-constant distance. Moreover, virtual coupling and splitting of trains when driving shall provide the flexibility for railway operators to adapt to the changing traffic demands and to use short platforms with long train compounds. Those future railway concepts are based on reliable T2T communication systems. Their design requires precise answers to two fundamental questions: How does the railway environment influence T2T signal propagation for frequencies above 1 GHz? How do high absolute speeds (up to 350 km/h) and high relative speeds (up to 700 km/h) influence the performance of T2T communication? Due to the lack of measurement campaigns on T2T signal propagation, the above questions have been open since long and motivated a new propagation measurement campaign. The focus of this campaign was on wagon-to-wagon (intra-consist) measurements within one high speed train (HST) and T2T measurements between two HSTs. These worldwide first high speed railway (HSR) channel sounding measurements took place in Italy on an HSR track between Naples and Rome in April 2016. The measurements were performed with a wide band channel sounder. Furthermore, the runs were used for evaluating an ITS-G5 system and a terrestrial trunked radio (TETRA) system under high speed conditions. The structure of this paper is as follows: we first give an overview of typical environmental aspects for HSRs in Section 2. In Section 3 we describe the measurement campaign in detail, i.e. the scenarios, manoeuvres and the measurement equipment. In Section 4, the ITS-G5 measurement data is analysed and split into the three main environments rural, sub-urban and tunnel. Finally, the log-distance path loss models for the mentioned environments are presented in Section 5. 2 Environmental aspects for high speed railways The environment along railways significantly impacts the terrestrial wireless propagation. It can be classified into distinct areas, unique scenarios, and railway-typical obstacles as listed in Table 1 (cf. [[3]] for more details). A train, travelling from city A to city B, passes through a random series of all three different areas. Table 1. Environments in railway traffic [[3]] Area Special scenario Obstacles urban curve cross bridge sub-urban tunnel noise barrier rural bridge/viaduct catenary — cutting signalling system — open field roof — — buildings — — vegetation (tree) The environment around HSRs varies less compared with other railways as HSRs usually connect big cities with each other. Between the cities, the railways are mainly located in rural or sub-urban areas. In the later one, the tracks are often shielded with noise barriers. Due to the high speed, the curve radius of the track is rather large, e.g. several kilometres, and gradients are very small. Hence, HSRs are mainly found in the following four scenarios: tunnels, viaducts, cuttings, and open fields. Only the first three affect the wireless propagation considerably different to free space propagation. Furthermore, no level crossings are employed for the safe operation of HSRs. Hence, crossover bridges for road traffic and other railway traffic are used. Most of the HSRs are electrified. Thus, catenaries and pylons are placed along the track. The pylons and supporting metal constructions for catenaries or the signalling system are very close to the antennas on the train. These obstacles reflect or scatter the radio waves and result in frequently appearing multi-path components (MPCs). Note that channel measurements and models for T2G links have been studied already in detail. For example, [[5]] surveys T2G channel measurements and models. To the best of the authors' knowledge, similar T2T communication research and measurement based channel models for HSR environments have not been published. 3 Measurement campaign We performed the worldwide first high speed T2T measurement campaign. The gaps on T2T communication models and related measurements are pointed out [[3]]. Within this measurement campaign, we covered intra-consist and T2T channel sounding measurements. Furthermore, we installed two existing communication systems and investigated their behaviour. First, an ITS-G5 system was setup for inside train and T2T communication. Second, the TETRA based train collision avoidance system (TrainCAS) was installed and used in HSR environments for the first time. In the following sections, each system is described and the different scenarios, manoeuvres, and setups are explained. 3.1 Scenarios and manoeuvres Together with our project partner Trenitalia, we chose the track between Naples and Rome as shown in Fig. 1. This double tracked HSR offers a large variety of environments listed in Table 1. The track length is 220 km and the maximum allowed speed is 300 km/h. Fig. 1Open in figure viewerPowerPoint HSR line Naples–Rome. Image by Google, Map Data 2016 NOAA, U.S. Navy, NGA, GEBCO Image Landsat (2015) The manoeuvres differ between the measurements with one train and with two trains in four consecutive nights: For inter-consist measurements with one train, we drove at a constant velocity of from the railway station Napoli Centrale in the direction of Rome. At waypoint PK36 (roughly 36 km before the Rome Central Station) we stopped and drove back to Naples with . For the T2T measurements, we drove from Napoli Centrale in the direction of Rome with low speed performing different manoeuvres. Latest at waypoint PK36 both trains stopped and performed the manoeuvres on the way back to Naples at high speed. The standard manoeuvre was to approach and overtake the Tx train by the Rx train. Afterwards, the Rx train reduced speed and fell behind the Tx train again. This manoeuvre sequence was repeated several times. In detail, the Tx train was driving at a constant speed of either 50 or 250 km/h at one track. The Rx train was driving with the variable speed of 30 to 70 km/h or 200 to 300 km/h on the parallel track. Opposing manoeuvres were performed two times. Once in a rural environment at high speed and once in the sub-urban environment at low speed. Other manoeuvres like one train is passing by at a still standing train or two trains are driving at a constant distance (e.g. like in a platoon) were performed as well. These manoeuvres were mostly done while rearranging for the main manoeuvres. 3.2 Measurement equipment and setup Trenitalia supported the measurement campaign with two Frecciarossa ETR 500 HSTs. This train series shown in Fig. 2 consists of two locomotives and 11 coaches with a total length of 330 m [[6]]. We equipped both trains with our measurement systems and railway specified antennas. For simplicity, in all measurements, and for all systems Train 1 was used as Tx and Train 2 as Rx as shown in Fig. 3. Fig. 2Open in figure viewerPowerPoint Trenitalia Frecciarossa ETR 500 HST Fig. 3Open in figure viewerPowerPoint Antenna setup for ITS-G5 measurements: Train 1 (top) and Train 2 (bottom) 3.2.1 Channel sounder We used the DLR-RUSK channel sounder as the main measurement system. The channel sounder was setup at a centre frequency of 5.2 GHz and a bandwidth of 120 MHz. A detailed description of this channel sounder is published in [[3]]. The time duration between the measurements of two consecutive channel impulse responses was set to and the maximum excess delay to . The channel sounder was set up for either intra-consist measurements in a single-input multiple-output mode in the first night of measurements or for T2T measurements in single-input single-output mode in the second, third and fourth night. An overview of all channel sounder measurements is given in Table 2; a detailed description can be found in [[7]]. The channel sounder data analysis is not presented in the following sections due to on-going work in the Roll2Rail project [[8]] and will be published at a later stage. Table 2. Channel sounder measurements and setup [[7]] Night Type Antenna GTx, W GAnt, dBi EIRP, dBm Distance, m 1 intra-consist omni 5 8 20 3, 29 3 T2T omni 40 8 33 variable 4 T2T direct 40 9 34 variable Table 3. ITS-G5 measurements and setup [[7]] Night Type Antenna GTx, W GAnt, dBi EIRP, dBm Distance, m 1 intra-consist omni 5 4 31 26, 52, 70 2 T2T omni 5 8 31 variable 3 T2T direct 5 9 31 variable 4 T2T omni 5 8 31 variable 3.2.2 ITS-G5 In addition to the channel sounder, we build up a second measurement system with ITS-G5 transceivers. ITS-G5 is a vehicular ad-hoc network communication standard based on IEEE 802.11p. The primary application is the connection of road vehicles with each other and with road side units. ITS systems operate in a dedicated frequency band at 5.9 GHz on seven 10 MHz channels, one control, four service channels and two channels reserved for future applications. The standard allows vehicles to transmit with a maximum equivalent isotropically radiated power (EIRP) of 33 dBm. The transmission technology uses orthogonal frequency division multiplexing. Different modulation and coding rates provide data rates from 3 up to 27 Mbits/s per channel. ITS-G5 enables a direct V2V communication up to a range of 3 km. Thus, it could support a variety of new applications in railway transportation like virtual coupling. Fig. 3 shows the setup for the ITS-G5 measurements with omni-directional antennas and Table 3 lists the corresponding antenna parameters. We installed two ITS-G5 Cohda MK5 transceiver modems and investigated intra-consist and T2T communication scenarios. In [[9]] we presented results of intra-consist (inside train) communication. In [[7]] we presented first results of the ITS-G5 T2T measurements. Simple one slope path loss models for open field and tunnel environment were presented in [[10]]. For all measurements, the radios were set up at the control channel at 5.9 GHz with a bandwidth of 10 MHz. All other settings are listed in Table 4 and set to ensure the most robust communication in order to assess the performance and to characterise the specific propagation channels in the various railway environments with their specific scattering and shadowing conditions. The Tx unit consists of the Cohda radio, a 5 W amplifier, connection cables and the railway antenna. The Tx chain was calibrated to achieve an EIRP of 31 dBm. On the Rx side, the Cohda radio was directly connected to the receiver antenna. End-to-end calibrations were performed before and after the measurement runs. Table 4. Cohda MK5 radio settings [[10]] Parameter Setting channel 180 carrier frequency 5.9 GHz bandwidth 10 MHz EIRP 31 dBm data rate 3 Mbit/s modulation binary phase shift keying coding rate 1/2 packet length 400 Byte repetition rate 100 Hz The ITS-G5 measurements were performed in parallel to the channel sounder measurements. In night two and four, the channel sounder was connected to directional antennas and ITS-G5 to omni-directional antennas. In night three, the systems were connected vice versa. The following data analysis and channel models are based on this ITS-G5 setup. 3.2.3 TrainCAS The TrainCAS is a special version of DLR's railway collision avoidance system (RCAS). The RCAS adopts the safety overlay concept as technology transfer from aircraft and maritime collision avoidance. As opposed to other traditional railway safety systems, this system does not require rely on any infrastructure, in contrast to the European train control system. TrainCAS exclusively relies on on-board technology [[2]]. The installed TrainCAS on-board units consist of three core technologies: a TETRA based T2T communication system, an accurate localisation system and a cooperative situation analysis and decision support system. As shown in Fig. 3, on each train a TETRA antenna was mounted and connected to the TrainCAS units. During the measurement campaign, the system's capabilities have been used to provide information to the train drivers and the mission responsibles about both trains involved in the measurement campaign. The online visualisation of both trains on a track map allowed a precise manoeuvre control for all measurement runs at different velocities. A detailed description of the TrainCAS system and the evaluation of the TETRA measurements can be found in [[11]]. 3.2.4 Antennas The different measurement systems demanded different antenna installations as shown in Fig. 3. Two kinds of omni-directional antennas were mounted on the roof of the trains. Due to safety regulations, only railway certified antennas could be mounted on the roof. A disadvantage of these antennas is the imperfect omni-directional behaviour as shown in Fig. 4. An advantage was the integrated global navigation satellite system (GNSS) antenna. Directional antennas were mounted behind the nose of the locomotives in 1.5 m height. Standard omni-directional antennas were used for the inside train measurements. In Table 5 all antennas and specifications are listed. Fig. 4Open in figure viewerPowerPoint Antenna pattern of the roof-mounted omni-directional antenna in the horizontal plane at 5.9 GHz Table 5. Antenna overview Antenna Type Frequency, GHz Location Amount Measurements Description SWA-0859/360/4/0/DFRX30_2 omni 5–6 coach roof 3 + 1 T2T, intra-consist railway specified SPA-2456/75/9/0/DF_1 directional 5–6 nose 1 + 1 T2T When the nose was closed, the antenna was in an upright position SWA-0459/360/4/25/DFRX30 omni, TETRA 0.3–0.5 locomotive roof 1 + 1 TrainCAS railway specified SWA-2459/360/7/20/V_1 omni 5–6 inside 2 ITS-G5 inside train standard indoor antenna 3.2.5 GNSS and inertial measurement unit (IMU) Each train was equipped with several GNSS receivers and IMUs to guarantee an accurate and reliable localisation during the measurements. The GNSS receivers recorded Global Positioning System, Galileo, and Global Navigation Satellite System signals. In particular, the pseudo-range, Doppler, carrier phase and carrier to noise ratio was stored. This data was post processed to calculate the train position and velocity. In tunnel sections, the GNSS receivers cannot be used for localisation due to the complete blockage of the satellite signals. Instead, in tunnels, the position and velocity are obtained from the IMUs. The IMUs measure the train accelerations and turn rates. With these measurements and the strap down algorithm published in [[12]], the attitude, position, and velocity of the train are estimated relative to the last GNSS position fix. Caused by the noise and biases on the IMU measurements, the accuracy of the position decreases with time and strongly depends on the IMU quality. For accurate localisation in long tunnels, the IMU-based approach is not sufficient and must be aided for example by the odometer measurements of the train. Nevertheless, the IMU provides important information while GNSS signals are received. Typically the IMU is combined with the GNSS measurements to detect outliers and to increase the update rate of the estimated position. Different GNSS receivers and IMU were used; the type and measurement rates of the different sensors are listed in Table 6. The Septentrio PolaRx4TR was used in combination with the channel sounder. Coordinates and timestamps were stored together with the channel sounder data; the raw data of the PolaRx4TR was stored for additional post processing. Table 6. Additional sensors [[7]] Name Sensor Update rate, Hz Septentrio PolaRx4TR GNSS 1 uBlox NEO-M8T GNSS 1 KVH 1750 IMU 100 Xsens MTi-G700 IMU 200 The uBlox NEO-M8T receivers were installed together with both IMU types in stand-alone measurement units. In each train, one localisation unit was placed in the locomotives and screwed to the floor frame. The GNSS signal was provided by the TETRA antennas. The GNSS and IMU raw data are stored with a GNSS time stamp on an SD card. The Cohda ITS-G5 units offer an integrated GNSS receiver. For omni-directional communication measurements, the units were connected to the GNSS port of the used measurement antennas. For directional measurements, additional GNSS antennas were placed inside the train behind the windows. The ITS-G5 measurement data was stored with position, speed and a time stamp on an SD card. 4 Data analysis The data analysis and following path loss models are based on the recorded received signal strength indication (RSSI) values of the ITS-G5 measurements. In night 4 the ITS-G5 equipment was connected to the omni-directional antennas on the roof. An RSSI value was only stored if the related transmitted packet could be received and decoded. One RSSI value is an average of the received power of one received packet. The Rx sensitivity was −98 dBm. Slow and fast overtaking manoeuvres in rural and tunnel environments, as well as a slow crossing in the sub-urban environment, were performed. The antenna pattern of the omni-directional antenna shown in Fig. 4 is taken into account. Hence, the differences in the measured received power for different train constellations could be compensated (either Train 1 or Train 2 in front). In [[7]] the impact of the antenna patterns is described, but not compensated. The following data analysis and path loss models are divided in the environments rural, sub-urban and tunnel. Figs. 5–7 show in the top part the measured received power and in the lower part the distance between transmitter and receiver train over time. Fig. 5Open in figure viewerPowerPoint Overtaking in a rural environment at high speed [[10]] Fig. 6Open in figure viewerPowerPoint Crossing in the sub-urban environment at low speed Fig. 7Open in figure viewerPowerPoint Slow crossing in the suburbs of Naples. Image by Google, Map Data 2017 NOAA, U.S. Navy, NGA, GEBCO Image Landsat (2017) 4.1 Rural The rural area is defined as an open field with straight and curved track sections. We assume line of sight (LOS) conditions. The manoeuvre recorded in Fig. 5 shows a platooning of both trains from 0 s to 200 s in the first phase. The distance varies between 320 and 400 m. Both trains accelerate almost synchronously from 100 to 250 km/h. In the second phase, from 200 to 270 s, Tx keeps its velocity constant at 250 km/h while Rx continues accelerating to 300 km/h. Thus, Rx approaches and overtakes Tx. The received power follows the free-space path loss (FSPL) model and additionally, deep fades are observed as well. 4.2 Sub-urban In the sub-urban area, we observed buildings and noise barriers close to the track. As shown in Fig. 8 straight and curved track sections occur. Hence, LOS and non-LOS (NLOS) conditions alternate accordingly. The recorded data for the sub-urban environment is shown in Fig. 6. The starting point and driving direction of the Tx train and Rx train are indicated by orange and green arrows in Fig. 8. The yellow rectangle indicates the crossing area. The distance plot shows the approach and separation of both trains from 1200 to 0 m and again to 950 m. The relative speed was . From 0 to 35 and 60 to 85 s strong fading effects can be observed. At these measurements, the distance was larger than 300 m. Shadowing effects at high distances can be explained by the factory buildings inside the track curve as shown in Fig. 8. Strong MPCs may arise from the surrounding buildings and close noise barriers. Fig. 8Open in figure viewerPowerPoint Overtaking in tunnels at high speed 4.3 Tunnel The overtaking manoeuvre took place in several consecutive straight and slightly curved tunnels as shown in Fig. 9. Not only LOS but also NLOS conditions need be assumed for different train constellations. The Rx train was driving at 150–200 km/h and the Tx train at 50–280 km/h. A detailed description of this manoeuvre is given in [[10]]. Fig. 9Open in figure viewerPowerPoint Overtaking in several tunnels. Image by Google, Map Data 2017 NOAA, U.S. Navy, NGA, GEBCO Image Landsat (2017) Three interesting environmental constellations could be identified in Fig. 7. The first constellation is related to the red, blue and magenta marked tunnels in Fig. 9. The received power for this multi-tunnel constellation is shown in Fig. 7 between 25 and 90 s. Due to the short distances between the tunnels, the received power level is up to 12 dB above FSPL, although the trains are in different tunnels. This effect can be observed at 230 s as well, but not as strong. The influence on the received power of the second train constellation can be found between 130 and 200 s in Fig. 7. At this time both trains drive inside the yellow marked 4 km long tunnel (see Fig. 9). With a distance of more than 2 km, we observed a gain caused by wave guiding effects of 15 dB in comparison with the FSPL model. Similar effects on the received power can be observed at 250 s, where both trains are in the brown tunnel. The third constellation is a mixed scenario with one train in a tunnel and the other train completely outside in open field. Diffraction effects on the tunnel portals cause deep fades as visible in Fig. 7 between 230 and 250 s. At this time one train was already in the brown tunnel and the other one between the green and brown tunnel. Hence, an additional attenuation of 15 to 20 dB could be observed. 5 Path loss models We fitted the measured RSSI values with common path loss models. For unobstructed cases and as a reference the FSPL model was calculated for the given link distances d as (1) From a certain link distance, all of our measurements show a mixture of LOS and NLOS situations with shadow fading. Hence, the log-distance path loss model was applied. In (2) the total path loss PL(d) is composed of a path loss PL(d0) at a fixed reference distance , the path loss exponent (PLE) n times the logarithm of the relation between the actual link distance d and and a Gaussian distributed vector (2) The shadow fading is modelled as Gaussian distribution function (3) with the standard deviation . Out of the data analysis done in Section 4, we establish path loss models for three different environments. In Figs. 10–12 the measured received signal over distance is plotted. The FSPL model is plotted always as a red solid line in each figure as the reference. One or two slope log-distance models are fitted to the received signals and plotted as well. The probability density of the shadow fading effects according to the log-distance models is plotted in Figs. 13–15. All model parameters are listed in Table 7. Fig. 10Open in figure viewerPowerPoint Log-distance path loss model for rural environment [[10]] Fig. 11Open in figure viewerPowerPoint Probability density of shadow fading in the rural environment [[10]] Fig. 12Open in figure viewerPowerPoint Two stage log-distance path loss model for sub-urban environment Fig. 13Open in figure viewerPowerPoint Probability density of shadow fading in sub-urban environment (a) Near region, (b) Far region Fig. 14Open in figure viewerPowerPoint Two stage log-distance path loss model for tunnel environments Fig. 15Open in figure viewerPowerPoint Probability density of shadow fading for tunnel environments (a) Near region tunnels, (b) Far region tunnels, (c) Far region mixed Table 7. Log-distance model parameters Region , m PL(d0), dB n , dB rural 1 47.86 2.14 4.22 sub-urban near 10 60.36 3 4.95 far 400 108.7 2.15 4.93 tunnels near tunnels 80 59.42 4.15 5.01 far tunnels 800 101.42 1.85 6.96 far mixed 500 106.83 2.3 5.39 5.1 Rural The received signal and the path loss models for the rural environment are plotted in Fig. 10; the log-distance model was presented in [[10]] and the effects described in detail. The proposed model slightly differs from FSPL. The higher PLE fits the path loss above 200 m better. Above distances of 300 m, we observed a mixture of LOS and NLOS situations caused by MPCs and shadowing effects. Hence, deeper fades can be observed in Fig. 10. These effects cause the spikes above the normal distribution fitting in Fig. 13. 5.2 Sub-urban The received power obtained in sub-urban environments was fitted with two slope models in Fig. 11. Similar to the approaches done for tunnels in [[13]–[15]], we introduce a near and far region. The use of two slope models in the near and far region is rather common for propagation modelling in tunnels. We use the two-slope approach for sub-urban environments since large buildings and long noise barriers close to the track behave similarly as a tunnel without a roof: a wave guiding effect cannot be observed, but strong MPCs occur and cause high attenuation in the near region. The near region is modelled with a PLE of 3; at 400 m the far region continues with an n = 2.15. For both regions, we plotted the probability density of the shadow fading in Fig. 14. For the near region, see Fig. 14a, the normal distribution fits well to the measured signal. Fig. 14b holds for the far region with a standard deviation of 4.93 dB. An overlap of LOS and shadowed LOS situations need to be assumed. The shadowed LOS components are responsible for the peaks above the normal distribution to the left, while the low extension to the right of the probability density can be related to the LOS situations. A similar behaviour was presented in [[16]] for V2V shadow fading models. 5.3 Tunnel In the same way, as for the sub-urban environment, we applied the two-slope model approach for tunnels. In Fig. 12, the received power is plotted in green, orange and blue dots. The green dots indicate the received power when both trains are in the same tunnel. The orange dots indicate the received power when the two trains are in two different but adjacent tunnels. In comparison with the FSPL model, a gain of 15–20 dB can be observed for both scenarios. Therefore, both scenarios are modelled jointly in a two-slope log-distance model. The near region up to 800 m is fitted with a PLE of 4.15 and indicated with a blue dashed line in Fig. 12. The reason for the high PLE is the high number of propagating MPCs. In the far region, most MPCs disappear and the LOS component dominates. Furthermore, the wave guiding effect causes a decrease of the PLE to 1.85 (see the black dotted line in Fig. 12). RSSI values up to −60 dBm could be measured at distances above 2 km. Nevertheless, especially for the far region above 800 m, large power variance occur which results in a high standard deviation of 6.96 dB as presented in Fig. 15b. The standard deviation of the near region model is 5.01 dB, see Fig. 15a. This path loss model aligns with presented multi-slope models for 400 MHz propagation in tunnels in [[17]]. The blue dots in Fig. 12 represent RSSI values where one train was in a tunnel and the other train completely outside in an open field environment. The diffraction and insertion effects at the tunnel portals cause significant losses. Nevertheless, in some measurements, LOS exists because RSSI as large as those provided by the FSPL model could be observed. This mixed environment is modelled with the magenta dotted-dashed line with a PLE of 2.3. Within this environment, the shadow fading of the received signal is normally distributed and varies with a standard deviation as shown in Fig. 15c. The parameters of all log-distance models for all environments are summarised in Table 7. 6 Conclusion In this study, we presented the worldwide first HSR measurement campaign for intra-consist and train-tot-rain (T2T) communications. In the first part of this publication, we explained the intention of this measurement campaign and gave an overview of the used measurement equipment and the setups. The DLR-RUSK channel sounder, an ITS-G5 communication system, and the TETRA based TrainCAS were used for T2T channel measurements under HSR conditions. Furthermore, TrainCAS was used for the manoeuvre control. The measurements were aligned with typical environments and manoeuvres for HSTs. The second part of this publication focused on the analysis and modelling of the ITS-G5 measurements. We classified the environments into rural, sub-urban and tunnel environments. The received power for the rural environment was modelled with a one slope log-distance path loss model. The resulting model behaves as expected near the FSPL model. For the sub-urban and tunnel measurements, we applied two slope log-distance path loss models. In sub-urban environments, the break point between near and far region was 400 m. In comparison with vehicle-to-vehicle road measurements, for HSRs the break point is considerably larger and the PLE behaves vice versa. This means, that in HSR sub-urban environments, the PLE for near region is larger than the PLE for the far region. To the best knowledge of the authors, this behaviour is already known for railway tunnels, but not for railway sub-urban environments. The log-distance model for tunnel environments shows similar behaviour as models in other publications. Similarly, as for sub-urban, the PLE for the near region was larger than the PLE for the far region. In the far region, the model can be applied to either the scenario where two trains are in one tunnel, as well as to the scenario where two trains are in adjacent tunnels. 7 Acknowledgments The authors are thankful for the support of the European Commission through the Roll2Rail project [[8]], one of the lighthouse projects of Shift2Rail [[18]] within the Horizon 2020 program. The Roll2Rail project has received funding from the European Union's Horizon 2020 research and innovation program under grant agreement no. 636032. 8 References [1]He, R., Ai, B., Wang, G., et al.: 'High-speed railway communications: from GSM-R to LTE-R', IEEE Veh. Technol. Mag., 2016, 11, (3), pp. 49– 58 [2]Lehner, A., Rico García, C., Strang, T.: 'On the performance of TETRA DMO short data service in railway VANETs', Wirel. Pers. Commun., 2013, 69, (4), pp. 1647– 1669 [3]Unterhuber, P., Pfletschinger, S., Sand, S., et al.: 'A survey of channel measurements and models for current and future railway communication systems', Mob. Inf. Syst., 2016, 2016, pp. 1– 14. Available at: http://www.hindawi.com/journals/misy/2016/7308604/ [4]Winter, J., Lehner, A., Polisky, E.: 'Electronic coupling of next generation trains'. Third Int. Conf. on Railway Technology: Research, Development and Maintenance', 2016, pp. 1– 17 [5]Wang, C.X., Ghazal, A., Ai, B., et al.: 'Channel measurements and models for high-speed train communication systems: a survey', IEEE Commun. Surv. Tutorials, 2016, 18, pp. 974– 987 [6] Trenitalia: ' ETR 500' ( Gruppo FS Italiane, 2016). Available at: http://www.trenitalia.com/ [7]Unterhuber, P., Sand, S., Soliman, M., et al.: 'Wide band propagation in train-to-train scenarios – measurement campaign and first results'. IEEE 11th European Conf. on Antennas and Propagation (EuCAP), 2017. pp. 3356– 3360 [8] Horizon2020: ' Roll2rail'. Roll2Rail Consortium, 2015. Available at: http://www.roll2rail.eu/ [9]Soliman, M., Unterhuber, P., Gera, D.: 'First analysis of inside train communication with ITS-G5 measurement data'. IEEE 2016 Int. Symp. on Wireless Communication Systems (ISWCS), 2016, pp. 451– 455 [10]Unterhuber, P., Soliman, M., Lehner, A.: ' ITS-G5 channel models for high speed train-to-train communication', in A. Pirovano, et al. (Ed.): ' communication technologies for vehicles' ( Springer International Publishing, Cham, 2017), pp. 72– 83 [11]Lehner, A., Strang, T., Unterhuber, P.: 'Train-to-train propagation at 450 MHz'. IEEE 11th European Conf. on Antennas and Propagation (EuCAP), 2017, pp. 2875– 2879 [12]Titterton, D., John, W.L.: ' Strapdown inertial navigation technology', IEEE Aerosp. Electron. Syst. Mag., 2005, 20, (7), pp. 33– 34, doi: 10.1109/MAES.2005.1499250 [13]Briso Rodriguez, C., Cruz, J.M., Alonso, J.I.: 'Measurements and modeling of distributed antenna systems in railway tunnels', IEEE Trans. Veh. Technol., 2007, 56, pp. 2870– 2879 [14]Guan, K., Zhong, Z., Ai, B., et al.: 'Propagation mechanism modelling in the near region of circular tunnels', IET Microw. Antennas Propag., 2012, 6, (3), pp. 355– 360 [15]Guan, K., Zhong, Z., Ai, B., et al.: 'Complete propagation model structure in-side tunnels', Prog. Electromagn. Res., 2013, 141, pp. 711– 726 [16]Abbas, T., Sjöberg, K., Karedal, J., et al.: 'A measurement based shadow fading model for vehicle-to-vehicle network simulations', Int. J. Antennas Propag., 2015, 2015, pp. 1– 12. Available at: https://www.hindawi.com/journals/ijap/2015/190607/ [17]Hrovat, A., Kandus, G., Javornik, T.: 'Four-slope channel model for path loss prediction in tunnels at 400 MHz', IET Microw. Antennas Propag., 2010, 4, (5), pp. 571– 582 [18] Horizon2020: 'Shift2rail'. Shift2Rail Joint Undertaking, 2015. Available at: http://www.shift2rail.org Citing Literature Volume12, Issue4March 2018Pages 492-500 FiguresReferencesRelatedInformation