Channel Measurements and Models For High-Speed Train Communication Systems A Survey

974 IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 18, NO.
2, SECOND QUARTER 2016
Channel Measurements and Models for High-Speed

Train Communication Systems: A Survey
Cheng-Xiang Wang, Senior Member, IEEE, Ammar Ghazal, Student Member, IEEE, Bo Ai, Senior Member, IEEE,
Yu Liu, and Pingzhi Fan, Fellow, IEEE
Abstract—The recent development of high-speed trains (HSTs) of their locations or speeds. To satisfy these demands, HST
as an emerging high mobility transportation system, and the grow- wireless communication systems have to overcome many chal-
ing demands of broadband services for HST users, introduce new lenges resulting from the high speed of the train that can easily
challenges to wireless communication systems for HSTs. Accurate
and efficient channel models considering both large-scale and non- exceed 250 km/h, such as fast handover, fast travel through
stationary small-scale fading characteristics are crucial for the diverse scenarios, and large Doppler spreads [11], [12] besides
design, performance evaluation, and parameter optimization of some challenges inherited from conventional trains such as high
HST wireless communication systems. However, the characteris- penetration losses, limited visibility in tunnels, and the harsh
tics of the underlying HST channels have not yet been sufficiently electromagnetic environment [13].
investigated. This paper first provides a comprehensive review of
the measurement campaigns conducted in different HST scenarios Since 1998, the Global System for Mobile Communication
and then addresses the recent advances in HST channel models. Railway (GSM-R) has widely been adopted as Europe stan-
Finally, key challenges of HST channel measurements and mod- dard for train communications and control. However, GSM-R
els are discussed and several research directions in this area are can only provide a data rate of up to 200 kbps [14], besides
outlined. the fact that it is mainly used for train control rather than pro-
Index Terms—High-speed train channels, channel measure- viding communications for train passengers [15]. Therefore,
ments, non-stationary channel models, statistical properties. GSM-R cannot meet the requirements for future high speed
data transmissions [16] and International Union of Railways
I. I NTRODUCTION has recommended that GSM-R has to be replaced by long-
term evolution-Railway (LTE-R) [17]–[23], which is a broad-
H IGH-MOBILITY scenarios, e.g., high-speed train (HST)
and vehicle-to-vehicle (V2V) scenarios, are expected to
be typical scenarios for the fifth generation (5G) wireless com-
band railway wireless communication system based on LTE-
Advanced (LTE-A) [24]. Nevertheless, both systems still adopt
munication systems [1]. Unlike V2V communication channels the conventional cellular architecture where mobile stations
that have been thoroughly investigated in the literature [2]–[10], (MSs) inside trains communicate directly with outdoor base sta-
a comprehensive study of HST communication channels is tions (BSs). Such an architecture leads to a spotty coverage and
still missing. With the rapid development of HSTs, an increas- high penetration losses of wireless signals traveling through the
ing volume of wireless communication data is required to be metal carriages of HSTs. In addition, the receiving signals at
transferred to train passengers. HST users demand high net- MSs on board will experience fast changing channels resulting
work capacity and reliable communication services regardless in high signaling overhead and high possibility of drop calls and
handover failure [25].
Manuscript received April 21, 2015; revised October 9, 2015; accepted The aforementioned problems can be mitigated by deploying
November 11, 2015. Date of publication December 17, 2015; date of current
version May 20, 2016. This work was supported in part by the EU FP7 QUICK
other cellular architectures, such as distributed antenna system
Project under Grant PIRSES-GA-2013-612652, EU H2020 5G Wireless Project (DAS) [26]–[28], coordinated multipoint (CoMP) [29], [30],
under Grant 641985, in part by the National 863 Project in 5G by Ministry mobile relay station (MRS) [31]–[34] (or mobile femtocell [1],
of Science and Technology in China under Grant 2014AA01A706, in part [35], [36]) technologies, or a combination of these architec-
by the National Natural Science Foundation of China under Grants 61210002
and 61222105, in part by the Natural Science Base Research Plan in Shaanxi tures, e.g., DAS with MRS [37] or CoMP with MRS [38]. In
Province of China under Grant 2015JM6320, in part by the National 973 a DAS, distributed antenna elements are connected to a BS via
Program under Grant 2012CB316100, and in part by 111 Project under Grant wires or fibers (radio over fibers (RoF)) [39], [40] to provide
111-2-14. considerable gain in coverage and capacity in comparison with
C.-X. Wang and A. Ghazal are with the Institute of Sensors, Signals,
and System, School of Engineering and Physical Sciences, Heriot-Watt the conventional cellular architecture. The spatially separated
University, Edinburgh EH14 4AS, U.K. (e-mail: [email protected]; antenna elements can be used to transmit the same signal at
[email protected]). different locations to provide spatial diversity against the fad-
B. Ai is with the State Key Laboratory of Rail Traffic Control
ing. Combined with spatial diversity, frequency reuse in the
and Safety, Beijing Jiaotong University, Beijing 100044, China (e-mail:
[email protected]). DAS is an effective technique to increase system capacity. The
Y. Liu is with the R&D Center on B4G Wireless Communication Networks, enhancement in spectral efficiency of DASs in comparison with
School of Information Science and Engineering, Shandong University, conventional systems was presented in [26]. In [27], the authors
Shandong 250100, China (e-mail: [email protected]).
P. Fan is with the Institute of Mobile Communications, Southwest Jiaotong
analyzed the deployment of DAS over HST communication
University, Chengdu, Sichuan 610031, China (e-mail: [email protected]). systems and some of the resulting problems such as the cov-
Digital Object Identifier 10.1109/COMST.2015.2508442 erage of the remote antenna units (RAUs) and echo channel
1553-877X © 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
Authorized licensed use limited to: P.S.R. Engineering College - Sivakasi. Downloaded on February 15,2024 at 08:14:59 UTC from IEEE Xplore. Restrictions apply.
WANG et al.: CHANNEL MEASUREMENTS AND MODELS FOR HIGH-SPEED TRAIN COMMUNICATION SYSTEMS 975
networks scenario in the WINNER II channel model [42] and

rural macro-cell (RMa) scenario in the IMT-A channel model
[41] have only considered a rural environment for HSTs, while
neglecting other HST scenarios. The aforementioned propa-
gation scenarios will be introduced and explained in detail in
Section II.
The features of HST channels, e.g., non-stationarity and
large Doppler shift, significantly differ from those of low-
mobility mobile cellular communication channels. Therefore,
many measurement campaigns have been conducted in the lit-
Fig. 1. A HST communication system deploying MRSs.
erature to understand the underlying physical phenomenon in
HST propagation environments. Accurate channel models that
are able to mimic key characteristics of wireless channels play
effect. In CoMP systems, the transmission of neighboring BSs an important role in designing and testing HST communica-
will be coordinated in the downlink while the received signals tion systems. Realistic and reliable large-scale fading channel
at the uplink will be jointly processed. This will reduce the models, i.e., path loss (PL) and shadow fading (SF) mod-
inter-cell interference and improve the cell edge throughput. els, are indispensable for efficient and trustworthy network
CoMP systems will also provide an enhanced channel capacity deployment and optimization. Small-scale fading channel mod-
by using the statistically independent properties of the channels els are crucial in physical layer design in order to develop and
resulting from the wide spatial separation of antenna elements. test different transmission schemes, such as diversity of trans-
Adopting mobile femtocell architecture in HST communica- mission/reception, error correction coding, interleaving, and
tion systems can be performed by deploying dedicated MRSs equalization algorithms. Inaccurate channel models may lead
on the surface of the train to extend the coverage of the out- to over-optimistic or over-pessimistic performance evaluation
door BS into train carriages. As a result, we will have two results that will result in misjudgments in product development.
channels: an outdoor channel between the BS and MRS, and Moreover, inaccurate channel models may lead to inaccurate
an indoor one between the MRS and an MS of a train pas- link budgets that will result in huge errors of the estimated max-
senger as illustrated in Fig. 1. In this case, the BS will mainly imum distance between adjacent BSs. Consequently, this will
communicate with the MRS at high data rates instead of com- cause poor coverage and increased drop calls due to failed han-
municating with large numbers of MSs directly. An MRS and dovers between BSs when the distance is underestimated, and
its associated MSs within a train carriage are all viewed as a unnecessary overlapped coverage area with unjustified instal-
single unit to the BS, while the MSs will see the relevant MRS lation and maintenance cost of the extra installed BSs when
as a regular BS. It follows that an MRS can perform a group the distance is overestimated [47]. In the literature, several
handover on behalf of all its associated MSs, which can greatly large-scale and small-scale fading HST channel models were
reduce the frequent handover burden of the HST system [36]. proposed. This article will focus on the recent advances in
Since the complexity of radio resource allocation (i.e., transmit HST channel measurements and modeling and their future
power, data rates, scheduling, power and frequency allocation, challenges.
and antenna selection) in a BS is related to the number of active The rest of this paper is organized as follows. In Section II, an
users [26], the radio resource management complexity in one overview of HST channel measurements is provided. The state-
BS will be reduced significantly when dealing with a “group of-the-art of HST channel models is presented in Section III.
of users” rather than individuals. This promising MRS tech- Future research directions in HST channel measurements and
nology has been adopted by IMT-Advanced (IMT-A) [41] and models are outlined in Section IV. Finally, concluding remarks
WINNER II [42] channel models. are highlighted in Section V.
Moreover, the transmitter (Tx) and receiver (Rx) of a HST
wireless communication system encounter different channel
conditions due to the difference of surrounding geographical II. HST C HANNEL M EASUREMENTS
environments. The HST environment can be generally classi- Special attention has been given to HST channel measure-
fied into the following main scenarios: open space, viaduct, ments in recent years. Due to the high speed of the train and
cutting, hilly terrain, tunnels, and stations. Considering some the hostile HST environments, conducting accurate channel
unique setup of the aforementioned scenarios and some other measurements for HST communication systems is challenging
special HST scenarios, HST environment can be further clas- and needs to address particular hardware and software require-
sified into 12 scenarios [43]. HSTs can operate across one or ments, e.g., robustness, scalability, hardware redundancy and
more of these scenarios during its travel. The propagation char- traceability [48]. Many measurement campaigns [49]–[93] for
acteristics change significantly with the change of environments different HST environments were presented in the literature.
and the distance between the Tx and Rx, even in the same Here, we will briefly review and classify the important mea-
terrain. Scenarios have a close relationship with channel mod- surements for HST communications according to the scenar-
eling and measurements. Most standard channel models in the ios, cellular architecture, measurements’ setup parameters (i.e.,
literature, like UMTS [44], COST 2100 [45], and IMT-2000 antenna configuration, carrier frequency, and bandwidth), and
[46], failed to introduce any of the HST scenarios. The moving measured channel statistics, as shown in Table I.
976 IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 18, NO. 2, SECOND QUARTER 2016
TABLE I
I MPORTANT HST C HANNEL M EASUREMENTS
A. HST Propagation Scenarios

HST environments can be roughly classified into the fol-
lowing 6 scenarios: open space, viaduct, cutting, hilly terrain,
tunnels, and stations.
1) In the open space scenario [52], also called plain scenario
[51], the Rx is moving at a very high speed in a rural area where
the BS antenna is much higher than the surroundings [54]. This
environment focuses on large cells and continuous coverage
where the link between the fixed Tx and moving Rx normally
has a dominant line-of-sight (LoS) component. However, after Fig. 2. Classification of HST scenarios.
a certain distance, called breakpoint distance, the impact of
the sparse scatterers will be noticed at the Rx represented by
non-LoS (NLoS) components. As a result, the slopes of the while slightly affected by environments [95]. Based on the
PL and Ricean K -factor will be noticeably changed at the geographic nature and the distribution/height of the surround-
breakpoint leading to dual-slope PL model [94]. It has been ing scatterers, the open scenarios can be further classified
proved that there is a strong link between the breakpoint dis- into rural [53], urban, and suburban scenarios as illustrated
tance and the antenna height. For a certain site, as the antenna in Fig. 2.
height decreases, the breakpoint moves closer to the Tx. This is 2) The viaduct scenario is very common for HSTs [55]–[71].
because a bigger Fresnel zone is intercepted by the ground, usu- The main purpose of viaducts is to ensure the smoothness of the
ally covered by vegetation, when the antenna height is lower. rail and the high speed of the train. In this scenario, the radio
Furthermore, due to the influences of different environments, reflection, scattering, and diffraction caused by nearby scatter-
slight variations in the breakpoint distance can be noticed in ers, e.g., trees and buildings, can also be reduced significantly.
different scenarios. Therefore, it can be concluded that the The viaduct height and relative BS height have great influence
breakpoint distance is mainly determined by the antenna height on the received signal. Because of the relatively high altitude
TABLE II
HST S CENARIOS
of the viaduct in comparison with the surrounding terrain, the length, size, and shape of the tunnels and the encountered
LoS component is dominant in this scenario. However, the spar- waveguide phenomena have significant impact on the commu-
sity of the scatterers in the environment around the viaduct nication channel. Because of the long limited space, bounding
will still influence the received signal at the Rx [59]. Based of tunnel, and poor smoothness of the interior wall, propa-
on the relative altitude between the scatterers and the viaduct, gation characteristics of signals in tunnels are quite different
this scenario can be further classified into high viaduct and low from other scenarios. To overcome the problem of the lim-
viaduct scenarios. In the former, most scatterers located within ited visibility encountered in tunnels and to design an optimal
50 m from the viaduct are lower than the surface of the viaduct wireless communication network, leaky feeders and DAS are
and therefore their impact on the propagation characteristics is often deployed. However, as HSTs may require long tunnels,
negligible. In the low viaduct scenario [71], [72], some of the the leaky feeder solution is very expensive especially at high
nearby scatterers are higher than the surface of the viaduct and operating frequencies and its maintenance is considerably com-
consequently they introduce rich reflections and scattering complex [96]. As a result, DAS is more practical [97]. It can provide
ponents that may result in a severe shadow fading and/or extra considerable gains in coverage and capacity, and provide spa-
pathloss [43]. tial diversity against the fading by using the antenna elements
3) The cutting scenario is another common scenario for at different locations. It also has advantages in future appli-
HST wireless communications [73]–[78]. It represents an envi- cations such as higher distance between repeaters and easy
ronment where the HST passes a U-shaped geographical cut maintenance after being opened.
surface between the hills. The cutting is widely used for HST 6) The stations scenario represents the railway facility where
construction to ensure the smoothness of the rail and help to HSTs stop regularly to load/unload passengers [83], [84]. HST
achieve a high speed of the train when passing through hills. stations can be classified according to their size or architecture.
The propagation of radio waveforms in this scenario is sig- Based on the size of the station, which reflects the estimated
nificantly affected by the steep walls on both sides. The LoS communication traffic, station scenario can be categorized into
component can be observed along the route of the HST in this small to medium size stations, large stations, and marshalling
scenario. Here, we can recognize between two cutting scenar- stations [43]. From the architecture perspective, which affects
ios: deep cutting if the receive antenna mounted on top of the the propagation characteristics inside the station, three HST sta-
train is lower than the upper eave of the cutting and low cut- tion scenarios can be recognized, i.e., open station, semi-closed
ting if the height of the upper eave is lower than the top of the station, and closed station [84] as illustrated in Fig. 2. Table II
receive antenna. briefly summarizes the description and key-parameters of the
4) In the hilly terrain scenario [79], [80], the surrounding aforementioned scenarios.
environment is densely scattered with objects distributed irreg- The aforementioned scenarios are the most encountered ones
ularly and non-uniformly. With high-altitude transmit antennas in HST environments. However, recent measurement cam-
and low-altitude obstacles, the LoS component is observable paigns have shed some light on other special HST scenarios
and can be detected along the entire railway. However, mul- such as crossing bridges [86]. Besides the previous “individual”
tipath components scattered/reflected from the surrounding scenarios, HSTs may encounter more than one scenario (the so-
obstacles will cause serious constructive or destructive effects called combination scenario [43]) in one cell. Two combination
on the received signal and therefore influence the channel’s scenarios are reported in the literature. The first one is a com-
fading characteristics. bination between tunnel and viaduct where viaducts are usually
5) The tunnel scenario represents an environment where used as transition between tunnels in mountain environments.
HST passes through tunnels [81], [82] with different lengths The frequent transition between tunnels and viaducts will
ranging from hundreds of meters to several kilometers. The increase the severity of fading at the transition points causing
a drop in the communication quality. The second combination will lead to a clear LoS and reduce the impact of these scatterers
is between cutting scenarios, i.e., deep and low cuttings, and on the received signal. The measurements of HST channels in
rural scenario. The frequent and fast transition between these cutting scenarios reported in [73], [74], [76], [77] have demon-
scenarios can degrade the quality of the communication link strated the impact of the cutting structure, i.e., the depth and
and makes signal prediction quite challenging. the widths of the top and bottom of the cutting, on the esti-
mated PLs. A shallow cutting, or low cutting, will lead to a
strong LoS condition while deep cutting will lead to a large
B. Channel Statistics
PL exponent due to the reflections from the cuttings’ slopes.
Channel statistics are essential for the analysis and design A comparison between the PL of cutting and viaduct scenarios
of a communication system. Most of HST measurement cam- was carried out in [73]. It was suggested that the propagation
paigns have concentrated on large-scale fading statistics, i.e., conditions in the cutting scenarios can be worse than those of
path loss (PL) and shadowing. The measurement campaign pre- viaduct ones because of the reflected and scattered components
sented in [49] studied the PL in HST channels when the Tx caused by the slopes of the cutting. It is important to note that
and Rx were located inside the same HST carriage and when such a conclusion is highly dependant on the dimensions of the
they were located in different carriages. The measured results studied viaduct and cutting, as we have highlighted the impact
showed that the waves traveling inside the same train carriage of those dimensions on the estimated PL earlier. On the other
are dominant compared to the ones reflected from scatterers hand, shadowing has generally been modeled as log-normal dis-
outside the HST due to the high penetration loss of wire- tributed in different HST scenarios. Various channel statistics
less signals traveling through the metal body of the carriages. studied in HST channel measurement campaigns are shown in
On the contrary, the waves reflected from outer scatterers are Table I.
dominant compared to the waves traveling inside the train car- The Ricean K -factor is a very important parameter in link
riages when the communication devices are located in different budget and channel modeling. Therefore, many papers pre-
carriages due to the high insulation between these carriages. sented the estimation of K -factors in different scenarios, e.g.,
In [50], the PL of indoor wideband HST channels was also open space [53], viaduct [55], [57], [58], [66]–[69], cutting
investigated using two different indoor Tx configurations, i.e., [73]–[77], and hilly terrain [79]. The previous discussions of the
omni-directional antenna mounted on the ceiling of the HST dominance of the LoS component, the breakpoint distance, and
and a planner antenna mounted on the wall of the carriage. the impact of the viaduct and cutting structure are also related
Measurements showed that the channel between the Tx plan- to the K -factor. For example, [69] showed how a higher value
ner antenna and Rx can suffer 10 dB greater loss compared of the viaduct height will lead to a higher value of the K -factor.
with the one between the Tx omni-directional antenna and Rx. In the same context, it showed that lower viaducts lead to more
The aforementioned results from both measurement campaigns surrounding scatterers, which results in an increase in the sever-
are very useful for the design of HSTs and measurement sce- ity of the fading and a considerable fluctuation of the K values.
narios. However, more measurements for indoor scenarios in Moreover, the measurement in [69] showed while the K -factor
HSTs are needed before confirming that these observations is a linear function of distance, the slopes of K values are dif-
are conclusive. PLs of HST channels in open space and hilly ferent before and after the breakpoint. Similar comprehensive
terrain scenarios were reported in [51], [53], [54] and [79], studies on K -factors of HST channels but in cutting scenar-
[98]. Measurement data reported in both hilly terrain scenarios were reported in [76], [99]. The analysis showed that wide
ios showed a breakpoint in the estimated PLs. A dominant and cuttings increase the possibility of dominant LoS components,
strong LoS component can be easily observed before the break- which leads to higher K -values. Distance-dependant linear K
point while the impact of scatterers starts and grows beyond models for different cutting dimensions before and after the
the breakpoint distance. The breakpoint distance depends on breakpoint distance were proposed in [76].
the clearance of the first Frensal zone and can be calculated In [65], [68], [69], [76], the spatial/temporal variations, e.g.,
based on the Tx and Rx antenna heights and the wavelength fade depth (FD), level-crossing rate (LCR), and average fade
of the transmitted signal [42]. Therefore, different breakpoint duration (AFD), were investigated. FD is a measure of varia-
distances were reported in the aforementioned hilly terrain mea- tion in the channel energy about its local mean due to small
surements, i.e., 778 m in [79] and 500 m in [98]. Since viaduct scale fading and it is calculated from the difference in signal
is a common HST scenario, the PL of HST viaduct channels has levels between 1% and 50%. Measurements in viaduct scenar-
thoroughly been studied in the literature, e.g., [55], [56], [59], ios have shown that FD is independent of the viaduct’s height
[66], [68], [69]. Most of these measurements highlighted the but is affected by the number and closeness of surrounding scat-
impact of the height of the viaduct and the relative height of the terers that are higher than the viaduct [65], [69]. LCR is defined
BS on the estimated PL. In general, there are two main obser- as the expected rate at which the received signal crosses a spec-
vations that can be concluded from the aforementioned viaduct ified level in a positive-going or negative-going direction, while
measurements. First, the higher the BS antenna, the smaller the AFD is defined as the average period of time for which the
PL exponent for a given viaduct height. Second, the viaduct received signal is below this specified level, i.e., threshold. LCR
reduces the severity of the channel fading. In other words, the and AFD were statistically modeled as functions of the struc-
higher the viaduct, the less fading severity. Both observations tural parameters of the viaduct and cutting scenarios in [69],
are physically meaningful considering that the increase of the [76]. The results showed that the severity of fading in viaduct
heights of the BS and the viaduct over the surrounding obstacles scenarios is greatly reduced compared with that in open space
TABLE III
PL AND S HADOW FADING M ODELS FOR HST C HANNELS
scenarios, since fewer reflected and scattered paths in viaduct C. Measurement’s Setup Parameters
scenarios are expected at the receiver which leads to smaller
values of LCR. Obstacles around the viaduct can cause minor Carrier Frequency and Bandwidth: most of the measurement
variations of the LCR values but have no significant impact on campaigns in the literature were conducted at the carrier fre-
the AFD. Cutting’s dimensions have also very minor impact quency of 930 MHz in GSM-R systems [54]–[58], [60]–[63],
on the AFD of the received signal, while surrounding obstacles [65], [68], [69], [73]–[76], [82], [87], [88]. Correspondingly,
and crossing bridges over the cutting have no influence on the all of the aforementioned measurements were for narrowband
LCR and AFD. Doppler behavior and angular information of channels with bandwidth of 200 kHz. Wideband channel mea-
HST channels in open space scenarios were analyzed in [52], surements with higher bandwidths, i.e., 10–100 MHz, and
while power delay profiles (PDPs) were investigated in [51], higher carrier frequencies, i.e., 2.1–5.2 GHz, were reported in
[53], [87], [91], [92]. In [96], a measurement was carried out in [49]–[53], [66], [67], [77]–[79], [81], [91]–[93].
a tunnel scenario and the signal propagation characteristics at Antenna Configuration: The majority of HST measurements
the breakpoint were discussed. campaigns so far have focused on single-input single-output
The stationarity interval, defined as the maximum time dura- (SISO) systems [49]–[51], [54]–[63], [65]–[69], [73]–[79],
tion over which the channel satisfies the wide sense stationary [81], [82], [87], [88], [91], [92]. Multiple-input multiple-
(WSS) condition, of HST channels was investigated in [88] output (MIMO) systems, where multiple antennas are equipped
based on measurements. It showed that conventional chan- at both ends, are essential for providing higher capacity to
nel models offered stationary intervals much larger than the meet the requirements of future high speed data transmis-
actual measured ones. In [100], the non-stationarity of a HST sions [101]. The channel measurement, particularly the MIMO
channel in a cutting scenario was investigated using a metric channel measurement at high moving speeds, remains to be
called non-stationarity index. The non-stationarity index was a challenging task. So far, only very few measurement cam-
defined as the distance between the auto-correlation of a real paigns were conducted using multiple antennas at either the
time-variant transfer function and the auto-correlation of this Tx, i.e., single-input multiple-output (SIMO) systems [52],
transfer function under the WSS assumption. The reported mea- [53], or Rx, i.e., multiple-input single-output (MISO) sys-
surement data showed that the non-stationarity index increases tems [52]. Hence, HST MIMO wideband channel measurement
when the Doppler frequency shift varies fast. In the future, more campaigns with carrier frequency and bandwidth larger than
channel statistics, especially those related to small-scale fading GSM-R ones are needed for future HST communication system
parameters, are necessary to be investigated in measurements. developments.
TABLE IV
I MPORTANT S MALL -S CALE FADING HST C HANNEL M ODELS
III. HST C HANNEL M ODELS environments and hence they can be modeled using existing
HST channel models in the literature can be categorized as indoor channel models [50]. Therefore, [41], [42], [109]–[111],
large-scale fading models [42], [50], [51], [56], [60]–[62], [66], [116] have focused on modeling the outdoor channel because of
the challenges that this channel faces due to the high velocity
[69], [73], [74], [77], [79] and small-scale fading models [24],
[41], [42], [102]–[116]. The state-of-the-art of HST channel of the Rx.
models has not been investigated yet. Therefore, we will first HST scenarios have been presented in details earlier in this
paper in Section II. While most of these scenarios can only
categorize PL models in Table III. In Table IV, the important
HST small-scale fading channel models are briefly reviewed be encountered in railway environments, open space scenario
and classified according to the modeling approach, scenario, is similar to the rural or urban scenarios that can be found in
conventional V2I or V2V communication systems. Therefore,
stationarity, antenna configuration, frequency selectivity (FS),
scatterer region, and cellular architecture. most of the current HST channel models, developed from V2I
and V2V channel models by taking into account the effect of
the high velocity of the Rx on the channel parameters, have
A. Large-Scale Fading Models been proposed for open space scenario [24], [41], [42], [102],
PL estimation is essential for wireless link budget compu- [103], [108]–[111], [114]. Channel models for tunnel, cutting,
tation and wireless network planning. PL and shadow fading and viaduct scenarios were studied in [106], [107], and [116].
channel models for various HST scenarios have been developed In summary, more HST channel models that consider other
based on measurement results conducted in the open literature cellular architectures, such as DAS, are needed in the future.
[3]–[14]. These PL models are typically expressed as In addition, more HST scenarios should be considered in
proposing future HST channel models.
P L (d) = A + 10n log10 (d) (1)
where d is the distance between the Tx and Rx in meters (m),
C. Modeling Approaches of HST Small-Scale Fading Models
n is the PL exponent, and A is the intercept. Note that the SF fol-
lows log-normal distributions, the standard deviation of which In terms of modeling approaches, the current HST chan-
for each model is given in Table III. nel models in the literature, presented in Table IV, can be
classified as deterministic [102]–[106] and stochastic channel
models. The latter can be further classified into geometry-
B. Cellular Architectures and Scenarios
based stochastic models (GBSMs) [41], [42], [107]–[111] and
As mentioned earlier, adopting conventional cellular archi- non-geometrical stochastic models (NGSMs) [114], [116], as
tecture in HST wireless communication systems may lead illustrated in Fig. 3.
to several problems in terms of providing reliable and fast 1) Deterministic Channel Models: Deterministic channel
communication to HST passengers. Therefore, other cellular models are usually based on the detailed description of spe-
architectures, such as DAS, CoMP, and MRS need to be con- cific propagation environment and antenna configuration. The
sidered. In the literature, most of the proposed channel models amplitudes, phases, and delays of the propagated waves are
have considered the conventional architecture where fixed BSs obtained using intensive simulations that incorporate details of
are installed on the track-side to provide wireless coverage propagation environments like roads, buildings, trees, houses,
to HST passengers inside carriages [24], [102], [103], [105]– etc. Therefore, deterministic models are physically meaningful
[108], [114]. By considering MRS solution, we will have two and potentially accurate. Geometry-based deterministic mod-
channels, outdoor channel between the BS and the MRS and an els (GBDMs) based on ray-tracing method were proposed in
indoor one between the MRS and train passengers. The proper- [103]–[106] to model HST propagation channels in different
ties of radio channels in the carriages resemble those of indoor HST scenarios. In [106], a three-dimensional (3D) ray-tracing
Fig. 4. A RS-GBSM with LoS and single-bounced rays for a HST cutting
scenario.
Fig. 3. Classification of non-stationary HST channel models.

applied to specific Tx, Rx, and scatterer geometries which are
predefined in a stochastic fashion according to certain probabil-
approach for wave propagation modeling in HST tunnels was ity distributions. Different types of GBSMs differ mainly in the
presented. The proposed model resulted in the complex channel proposed scatterer distributions. Based on the position of the
impulse response that incorporates channel information, e.g., effective scatterers, GBSMs can be further classified as regular-
the wave-guide effect observed in tunnels and the impact of shaped GBSMs (RS-GBSM) such as one-ring [108], two-ring,
another train passing in the opposite direction on the Doppler and ellipse models [109]–[111], and irregular shaped GBSMs
shift and time delay. The authors in [104] adopted a similar (IS-GBSMs) [41], [42], [107].
approach to model HST channels in various scenarios. Both RS-GBSMs assume that all the effective scatterers are placed
[106] and [104] used measurement results to verify the pro- on regular shapes and therefore, different RS-GBSMs have dif-
posed channel models. Another HST channel model based ferent shapes of scatterer distributions, e.g., one-ring, two-ring,
on 3D ray-tracing approach was presented in [103] to ana- and ellipses for two-dimensional (2D) models and one sphere,
lyze channel characteristics, e.g., the FS and time-variance two-sphere, and elliptic-cylinders for 3D ones. RS-GBSMs
(Doppler spread). The objects, e.g., trees, buildings, or barri- often result in closed-form solutions or at least mathemati-
ers, on both sides of the railway track were modeled using cally tractable formulas. The generalized principle of designing
rectangular boxes, the dimensions of which were statistically RS-GBSMs follows the following steps. First, a geometri-
generated. Since the propagation characteristics of electro- cal model is adopted assuming that scatterers are located on
magnetic (EM) waves in tunnels are significantly different regular shapes. Then, a stochastic reference model with an infi-
from those in other HST environments, a multi-mode waveg- nite number of scatterers is developed based on the adopted
uide channel model was proposed in [117]. The proposed geometrical model. However, the reference model cannot be
model, which is a hybrid model that combines the geometri- used for simulations and therefore a corresponding simula-
cal optical model and waveguide model, can characterize the tion model with a finite number of effective scatterers is
wave propagation both in near and far regions of the source. needed. The parameters of the simulation model are com-
However, the aforementioned model failed to discuss the far puted by using proper parameter computation methods, e.g., the
LoS (FLOS) phenomena observed inside tunnels [118] or pro- extended method of exact Doppler spread (EMEDS), modified
vide a mechanism to determine the breakpoint for different method of equal area (MMEA) [120], or the L p -Norm Method
propagation regions in tunnels [13]. A GBDM based on ran- (LPNM) [121]. In [108], a one-ring RS-GBSM was proposed
dom propagation-graph was proposed in [102] to characterize to model HST channels in open space scenarios. The scatterers
time-variant HST channels in open space scenarios. Similar were assumed to be distributed on a ring around the MS where
to ray-tracing method, propagation-graph can predict channel different PDFs of the scatterers were analyzed. Considering the
impulse responses by a thorough search of propagation paths narrowband GSM-R for a HST communication system, a 3D
that connect the Tx and Rx. This modeling approach can be one-sphere RS-GBSM was proposed in [107] for open space
performed by considering the geometry of the simulated envi- scenarios. The proposed model used the Von Mises distribution
ronments, e.g., the distribution, mobility, and visibility of the to describe the azimuth angles and the space-time (ST) cross-
scatterers. Despite their high accuracy, GBDMs require detailed correlation function (CCF) was derived. However, both of the
descriptions of the propagation environments and extensive aforementioned models assumed that the HST channel satisfies
computational resources to be implemented. To avoid the high the WSS condition that has been proved incorrect by measure-
complexity of implementing GBDMs while maintaining suffi- ments [52]. To fill this gap, non-stationary RS-GBSMs were
cient accuracy, semi-deterministic models for HST viaduct and proposed in [109]–[112] for wideband MIMO HST channels
cutting scenarios were proposed in [119]. However, the pro- considering the deployment of MRS on the top of the train.
posed models only considered large-scale fading and neglected Fig. 4 illustrates the proposed RS-GBSMs, which consist of
the effect of small-scale fading parameters on the received multiple confocal ellipses with single-bounced rays and the LoS
signal. component.
2) GBSMs: In GBSMs, the impulse responses of HST The model was first introduced in [109], [110], where it con-
channels are characterized by the law of wave propagation sidered the distance between the Tx and Rx as time-varying
to capture the non-stationarity of the HST channel. Then, IV. R ESEARCH D IRECTIONS IN HST C HANNEL
the model was further developed in [111] by considering M EASUREMENTS AND M ODELS
other time-varying model parameters, i.e., angles of departure
In this section, we will discuss important research directions
(AoDs) and angles of arrival (AoAs). By adopting some key that can be considered as guidelines for conducting future HST
scenario-specific channel parameters, this model was further measurement campaigns and developing more realistic HST
extended in [112] to be applicable to the three most common
channel models.
HST scenarios, i.e., open-space, viaduct, and cutting scenar-
ios [47], and hence is the first generic HST channel model.
To demonstrate its applicability, the proposed generic non- A. Nonstationarity of HST Channels
stationary HST channel model was verified by measurements
in terms of stationary time for the open space scenario and the Measurements in the literature have demonstrated that HST
autocorrelation function (ACF), LCR, and stationary distance channels are non-stationary since the stationary conditions,
for the viaduct and cutting scenarios [112]. measured by stationary interval, retain to a very short period
IS-GBSMs place the effective scatterers with predefined of time in comparison with other types of channels, e.g., V2I
properties at random locations with certain statistical distribu- and V2V channels [88]. This is mainly caused by the very high
tions usually obtained/approximated from measurements [122]. speed of the trains and the encountered changes in surrounding
Unlike RS-GBSMs, the random locations of the scatterers do areas. Although the non-stationarity of HST channels has been
not form regular shapes and the signal contributions of the implicitly considered in GBDMs [102]–[104], [106], but these
effective scatterers are determined from a greatly-simplified models are mainly site-specific and cannot be easily generalized
ray-tracing method and finally the total signal is summed up to a variety of scenarios. The non-stationarity feature of HST
to obtain the complex impulse response. IS-GBSMs for HST channels has been considered in the NGSM proposed in [114]
channels were introduced in the RMa scenario in WINNER II by implementing the birth-death process to simulate the appear-
[42] channel model and the moving networks scenario in IMT- ance and disappearance of the scatterers, and in RS-GBSMs in
A channel model [41]. In both cases, the train speed can be up [109]–[112] by providing the time-variant functions of several
to 350 km/h and the MRS technology is employed. In [107], channel model parameters, i.e., angular parameters, Doppler
an IS-GBSM was proposed for HST channels in cutting scenar- frequency, Ricean K -factor, and the distance between the Tx
ios assuming the scatterers to be uniformly distributed on the and Rx. However, verifying the proposed models by real-field
surface of the two slopes of the cutting. However, the aforemen- measurements was only performed in [112] and therefore more
tioned channel models have neglected the non-stationarity of comprehensive investigations are required to validate the accu-
HST channels and assumed that the WSS assumption can still racy of those models. Future non-stationary channel models
be applied. Moreover, GBSMs are very complex for upper-layer should consider more time-variant model parameters, such as
protocol design and performance analysis and less complex cluster powers and delays, and investigate the effect of the drift
channel models are preferred. of scatterers into different delay taps on the non-stationarity
3) NGSMs: NGSMs characterize physical parameters of a of HST channels and the resulting correlation between
HST propagation channel in a completely stochastic manner these taps.
by providing their underlying probability distribution func-
tions without assuming an underlying geometry. An NGSM
B. Statistical Properties of HST Channels
based on finite-state Markov chains for HST wireless com-
munication channels was proposed in [114]. The proposed Investigating the statistical properties of HST channels is
model is able to capture the characteristics of time-varying essential for understanding and analyzing HST communication
HST wireless channels by using Markov chains to track the systems. In Table I, several channel statistics obtained from
channel state variation at different received signal-to-noise ratio measurements were presented. However, most of proposed
(SNR) intervals. However, the model has not been verified by HST channel models in the literature have failed to provide
using real-field measurements and thus deserves more inves- the corresponding theoretical analysis. In [107], the ST CCF
tigation. The authors in [116] followed a similar approach to was derived based on the proposed stationary narrowband HST
model the dynamic evolution of multi-path components, i.e., channel model. In [111], a novel theoretical framework that
birth-death process, using a four-state Markov chain model. characterizes non-stationary mobile fading channels in terms of
The four proposed states are no birth/death, births only, deaths their system functions and correlation functions was proposed.
only, and both births and deaths. The transition matrix of the Based on this theoretical framework, different time-variant
birth-death process was calculated based on the measurement statistical properties of the RS-GBSMs in [111], [112] were
presented in [66]. Based on measurement of HST channels in derived, i.e., time-variant space CCFs, time-variant ACFs, time-
viaduct and cutting scenarios, a finite-state Markov channel was variant space-Doppler (SD) power spectrum densities (PSDs),
also proposed in [123]. Simulation results showed that Ricean local scattering functions (LSFs) [111], and LCRs [112].
distribution can well characterize the measured amplitude of It is highly desirable to investigate the statistical properties
the small-scale fading in both HST scenarios and an NGSM of other HST channel models and further develop the afore-
can effectively capture the dynamic nature of the fast fading in mentioned theoretical framework to include more statistical
HST channels. properties.
C. HST Scenarios obstacles and accidents via wireless communication links. A

HST scenarios were classified and thoroughly explained in HST-to-HST communication model based on multihop and
Section II of this article. During its travel and due to its high cooperation was proposed in [130]. In the proposed model,
velocity, HST runs across diverse scenarios so rapidly that a a source train uses trains on other tracks as relays to trans-
single model is incapable of capturing accurately the varia- mit signals to the destination train on the same track. Based
tions of HST channels. While most standard channel models did on proposed HST-to-HST channel model, the bit error rate
not consider any of the HST scenarios, IMT-A channel model, (BER) performance was investigated in [130] using the sub-
WINNER II channel model, and most of the non-standard HST urban scenario of COST 207 channel model and the outage
channel models were proposed for open space scenarios only. capacity was analyzed in [131] using Nakagami-m channel
Due to the unique feature of tunnels, the propagation character- model. Despite its importance as a safety measure to avoid acci-
istics of signals inside tunnels are different from those of other dents, the development of HST-to-HST channel models is still
HST scenarios. Conventional channel modeling techniques in its preliminary phase and further investigations are required.
suitable for other HST scenarios are not directly applicable
to tunnel scenarios. Moreover, research on channel modeling F. System Performance
inside tunnels is yet to solve main problems such as the accu-
rate characterization of multimode waveguide propagation and Investigating the performance of HST communication sys-
the determination of breakpoint for different propagation mech- tems is the basis for system design and network planning. In
anisms inside tunnels. In addition, the research on 3D channel [33], the HST communication system performance was inves-
models for tunnels is still in its very early stages and accurate tigated using data throughput to evaluate a seamless dual-link
generic channel models that can be applied to different types handover scheme. Another handover scheme was proposed in
of tunnel channels are still missing in the literature. Therefore, [34] and the system performance was evaluated by tracking
channel characterization and modeling for tunnel scenarios are the changes of throughput and signal-to-noise-and-interference
still a quite challenging topic and need to be further investi- ratio (SINR) over the time. The changes of SINR over the
gated. Moreover, it is essential that future HST channel models HST velocity were investigated in [132] to evaluate a trans-
consider other scenarios such as station scenarios that also mit beamforming algorithm proposed for canceling the inter-
have their unique channel characteristics, and preferably take channel interference (ICI) in HST communication systems.
into account the impact of diverse scenarios on HST channel The performance of HST communication system that imple-
models. ments beamforming technique was also evaluated in [133]
using measured throughput, SINR, and received signal strength
indicator level. The deployment of DAS in HST communica-
D. 3D HST Channel Models tion systems was evaluated in [26] by using spectrum efficiency
as a system performance metric. In [134], BER was used
Apart from the GBDMs that use 3D ray-tracing tool to model to evaluate a proposed radio resource allocation scheme for
HST channels [103]–[106], HST channel models were gener- orthogonal frequency-division multiple access (OFDMA) HST
ally proposed assuming that propagation waves are traveling in systems. The BER performance of a HST communication
two dimensions and therefore ignore the impact of the elevation system was also investigated in [135] where beamforming
angle on channel statistics. In reality, radio waves propagate and Alamouti combined downlink transmission schemes were
in three dimensions and scatterers are disperse in elevation, proposed. Mobile broadband performance experienced from
i.e., vertical plane, as well as in azimuth, i.e., horizontal plane. regional HSTs was investigated in [136] by monitoring the fluc-
Recently, the 3GPP has developed a 3D channel model in urban tuation of system throughput caused by the varying distance
microcell and urban macrocell scenarios following the frame- between the BS and HST, multi-path fading, and co-channel
work of WINNER II channel model [124]. The proposed 3D interference conditions. A temporal proportional fair power
3GPP channel model introduced the zenith AoD and zenith allocation scheme for HST wireless communication systems
AoA that are modeled by inverse Laplacian functions [125]. was proposed in [137]. The proposed scheme was designed
The 3D extensions of SCM and the WINNERII/WINNER+ to achieve a trade-off between power efficiency and fairness
channel models were proposed in [126] and [127], respec- along the time. HST channel capacity was analyzed in [138]
tively, and an extension of the IMT-A channel model to the to study the impact of different antenna array configurations on
elevation plane was proposed in [128], [129]. However, none MIMO HST communication systems. In [139], the BER per-
of the aforementioned channel models considered any of the formance of spatial modulation systems was studied using the
HST scenarios. Thus, 3D channel measurements and models proposed non-stationary HST MIMO channel model in [111]
are necessary, especially when the HST is close to the BS where with different HST scenarios. It was shown that the correla-
considering elevation angles can demonstrate the impact of the tion between sub-channels, inter-symbol-interference, Doppler
waves reflected from ground on the received signal. shift, and channel estimation errors are the main factors that
affect the BER performance of SM systems under the HST
channel model. More comprehensive system performance anal-
E. HST-to-HST Communications
ysis that evaluates other schemes and considers more system
HST-to-HST communication has been proposed to enable performance indicators, e.g., capacity and quality of service
HSTs exchange controlling and traffic information such as road (QoS), is required in the future.
V. C ONCLUDING R EMARKS [16] K. Kastell, S. Bug, A. Nazarov, and R. Jakoby, “Improvements in rail-
way communication via GSM-R,” in Proc. IEEE Veh. Technol. Conf.
This article has provided a survey of HST channels in terms (VTC’06-Spring), Melbourne, VIC, Australia, May 2006, pp. 3026–
of conducted measurements and channel models. We have 3030.
[17] J. Calle-Sánchez, M. Molina-García, J. I. Alonso, and A. Fernández-
classified HST channel measurements according to scenarios, Durán, “Long term evolution in high speed railway environments:
carrier frequencies, bandwidths, measured channels, antenna Feasibility and challenges,” Bell Labs Tech. J., vol. 18, no. 2, pp. 237–
configurations, train speeds, and channel statistics. We have 253, Sep. 2013.
[18] K. Guan, Z. Zhong, and B. Ai, “Assessment of LTE-R using high
also presented various HST large-scale fading channel mod- speed railway channel model,” in Proc. 3rd Int. Conf. Commun. Mobile
els in the literature. Then, we have classified HST small-scale Comput. (CMC’11), Qingdao, China, Apr. 2011, pp. 461–464.
fading channel models based on their modeling approaches, [19] J. Rodriguez-Pineiro et al., “LTE downlink performance in high speed
trains,” in Proc. IEEE Veh. Technol. Conf. (VTC’15-Spring), Glasgow,
scenarios, stationarity, FS, and cellular architecture. Finally, Scotland, May 2015, pp. 1–5.
we have highlighted some research directions in HST channel [20] S. Zhang, M. Wu, and X. Lin, “Results and analysis based on a
measurements and modeling. The discussions here will hope- novel LTE-R channel measurement at 2.6 GHz,” in Proc. 22nd Int.
Conf. Telecommun. (ICT’15), Sydney, NSW, Australia, Apr. 2015,
fully shed lights on recent challenges that face HST channels pp. 403–407.
and help in conducting future measurement campaigns and [21] M. Cheng, X. Fang, and W. Luo, “Beamforming and positioning-
developing more realistic HST channel models. assisted handover scheme for long-term evolution system in high-speed
railway,” IET Commun., vol. 6, no. 15, pp. 2335–2340, Oct. 2012.
[22] M. Alasali and C. Beckman, “LTE MIMO performance measurements
on trains,” in Proc. 7th Eur. Conf. Antennas Propag. (EuCAP’13),
R EFERENCES Gothenburgh, Sweden, Apr. 2013, pp. 2327–2330.
[23] F. J. Martin-Vega, I. M. Delgado-Luque, F. Blanquez-Casado,
[1] C.-X. Wang et al., “Cellular architecture and key technologies for 5G G. Gomez, M. C. Aguayo-Torres, and J. T. Entrambasaguas, “LTE per-
wireless communication networks,” IEEE Commun. Mag., vol. 52, no. 2, formance over high speed railway channel,” in Proc. IEEE Veh. Technol.
pp. 122–130, Feb. 2014. Conf. (VTC’13-Fall), Las Vegas, NV, USA, Sep. 2013, pp. 1–5.
[2] A. F. Molisch, F. Tufvesson, J. Karedal, and C. F. Mecklenbrauker, [24] 3GPP, “3rd Generation partnership project; Technical specification
“A survey on vehicle-to-vehicle propagation channels,” IEEE Wireless group radio access network; Evolved universal terrestrial radio access
Commun., vol. 16, no. 6, pp. 12–22, Dec. 2009. (E-UTRA); Base Station (BS) radio transmission and reception (Release
[3] C.-X. Wang, X. Cheng, and D. I. Laurenson, “Vehicle-to-vehicle chan- 11),” TS36.104, V11.3.1, Jan. 2013.
nel modeling and measurements: Recent advances and future chal- [25] K. Yamada, Y. Sakai, T. Suzuki, Y. Kawahara, T. Asami, and H. Aida,
lenges,” IEEE Commun. Mag., vol. 47, no. 11, pp. 96–103, Nov. “A communication system with a fast handover under a high speed
2009. mobile environment,” in Proc. IEEE Veh. Technol. Conf. (VTC’10-Fall),
[4] X. Cheng, C.-X. Wang, D. I. Laurenson, S. Salous, and A. V. Vasilakos, Ottawa, ON, Canada, Sep. 2010, pp. 1–5.
“An adaptive geometry-based stochastic model for non-isotropic MIMO [26] J. Wang, H. Zhu, and N. J. Gomes, “Distributed antenna systems
mobile-to-mobile channels,” IEEE Trans. Wireless Commun., vol. 8, for mobile communications in high speed trains,” IEEE J. Sel. Areas
no. 9, pp. 4824–4835, Sep. 2009. Commun., vol. 30, no. 4, pp. 675–683, May 2012.
[5] X. Cheng et al., “Cooperative MIMO channel modeling and multi-link [27] H.-A. Hou and H.-H. Wang, “Analysis of distributed antenna sys-
spatial correlation properties,” IEEE J. Sel. Areas Commun., vol. 30, tem over high-speed railway communication,” in Proc. IEEE 23rd Int.
no. 2, pp. 388–396, Feb. 2012. Symp. Pers. Indoor Mobile Radio Commun. (PIMRC’12), Sydney, NSW,
[6] X. Cheng, Q. Yao, M. Wen, C.-X. Wang, L. Song, and B. Jiao, Australia, Sep. 2012, pp. 1300–1305.
“Wideband channel modeling and ICI cancellation for vehicle-to- [28] C. Yang, L. Lu, C. Di, and X. Fang, “An on-vehicle dual-antenna han-
vehicle communication systems,” IEEE J. Sel. Areas Commun., vol. 31, dover scheme for high-speed railway distributed antenna system,” in
no. 9, pp. 434–448, Sep. 2013. Proc. IEEE 60th Int. Conf. Wireless Commun. Netw. Mobile Comput.
[7] Y. Yuan, C.-X. Wang, X. Cheng, B. Ai, and D. I. Laurenson, “Novel (WiCOM’10), Chengdu, China, Sep. 2010, pp. 1–4.
3D geometry-based stochastic models for non-isotropic MIMO vehicle- [29] D. Lee et al., “Coordinated multipoint transmission and reception
to-vehicle channels,” IEEE Trans. Wireless Commun., vol. 13, no. 1, in LTE-advanced: Deployment scenarios and operational challenges,”
pp. 298–309, Jan. 2014. IEEE Commun. Mag., vol. 50, no. 2, pp. 148–155, Feb. 2012.
[8] X. Cheng, C.-X. Wang, B. Ai, and H. Aggoune, “Envelope level cross- [30] C.-X. Wang, X. Hong, X. Ge, X. Cheng, G. Zhang, and J. S. Thompson,
ing rate and average fade duration of non-isotropic vehicle-to-vehicle “Cooperative MIMO channel models: A survey,” IEEE Commun. Mag.,
Ricean fading channels,” IEEE Trans. Intell. Transp. Syst., vol. 15, no. 1, vol. 48, no. 2, pp. 80–87, Feb. 2010.
pp. 62–72, Feb. 2014. [31] Y. Zhou, Z. Pan, J. Hu, J. Shi, and X. Mo, “Broadband wireless commu-
[9] Y. Yuan, C.-X. Wang, Y. He, M. M. Alwakeel, and H. Aggoune, “3D nications on high speed trains,” in Proc. IEEE 20th Annu. Wireless Opt.
wideband non-stationary geometry-based stochastic models for non- Commun. Conf. (WOCC’11), Newark, NJ, USA, Apr. 2011, pp. 1–6.
isotropic MIMO vehicle-to-vehicle channels,” IEEE Trans. Wireless [32] L. Tian, Y. Zhou, J. Li, Y. Huang, J. Shi, and J. Zhou, “A novel han-
Commun., vol. 14, no. 2, pp. 6883–6895, Dec. 2015. dover scheme for seamless wireless connectivity in high-speed rail,”
[10] Y. Fu et al., “BER performance of spatial modulation systems under in Proc. IEEE 7th Int. Conf. Wireless Mobile Comput. Netw. Commun.
3D V2V MIMO channel models,” IEEE Trans. Veh. Technol., 2015, (WiMob’11), Wuhan, China, Oct. 2011, pp. 230–236.
submitted for publication. [33] L. Tian, J. Li, Y. Huang, J. Shi, and J. Zhou, “Seamless dual-link han-
[11] F. Abrishamkar and J. Irvine, “Comparison of current solutions for the dover scheme in broadband wireless communication systems for high-
provision of voice services to passengers on high speed trains,” in Proc. speed rail,” IEEE J. Sel. Areas Commun., vol. 30, no. 4, pp. 708–718,
IEEE Veh. Technol. Conf. (VTC’00-Fall), Boston, MA, USA, Sep. 2000, May 2012.
pp. 1498–1505. [34] O. B. Karimi, L. J. Liu, and C. Wang, “Seamless wireless connectiv-
[12] D. T. Fokum and V. S. Frost, “A survey on methods for broadband ity for multimedia services in high speed trains,” IEEE J. Sel. Areas
Internet access on trains,” IEEE Commun. Surveys Tuts., vol. 12, no. 2, Commun., vol. 30, no. 4, pp. 729–739, May 2012.
pp. 171–185, Apr. 2010. [35] F. Haider et al., “Spectral efficiency analysis of mobile femtocell based
[13] B. Ai et al., “Challenges toward wireless communications for high- cellular systems,” in Proc. IEEE 13th Int. Conf. Commun. Technol.
speed railway,” IEEE Trans. Intell. Transp. Syst., vol. 15, no. 5, (ICCT’11), Jinan, China, Sep. 2011, pp. 347–351.
pp. 2143–2158, Oct. 2014. [36] F. Haider, C.-X. Wang, B. Ai, H. Haas, and E. Hepsaydir, “Spectral-
[14] M. Goller, “Application of GSM in high speed trains: Measurements and energy efficiency trade-off of cellular systems with mobile femtocell
simulations,” in Proc. IEE Colloq. Radiocommun. Transp., May 1995, deployment,” IEEE Trans. Veh. Technol., 2015, submitted for publica-
pp. 1–7. tion.
[15] C. Briso, C. Cortes, F. J. Arques, and J. I. Alonso, “Requirements of [37] X. Qian, H. Wu, and J. Meng, “A dual-antenna and mobile relay station
GSM technology for the control of high speed trains,” in Proc. IEEE based handover in distributed antenna system for high-speed railway,”
13th IEEE Int. Symp. Pers. Indoor Mobile Radio Commun. (PIMRC’02), in Proc. IEEE 7th Int. Conf. Innov. Mobile Internet Serv. Ubiq. Comput.
Lisbon, Portugal, Sep. 2002, pp. 792–793. (IMIS’13), Taichung, Taiwan, Jul. 2013, pp. 585–590.
[38] W. Luo, R. Zhang, and X. Fang, “A CoMP soft handover scheme for [61] R. He, Z. Zhong, B. Ai, L. Xiong, and J. Ding, “The effect of reference
LTE systems in high speed railway,” EURASIP J. Wireless Commun. distance on path loss prediction based on the measurements in high-
Netw., vol. 2012, 9 p., 2012, doi: 10.1186/1687-1499-2012-196. speed railway viaduct scenarios,” in Proc. CHINACOM, Harbin, China,
[39] B. Lannoo, D. Colle, M. Pickavet, and P. Demeester, “Radio-over-fiber- Aug. 2011, pp. 1201–1205.
based solution to provide broadband Internet access to train passengers,” [62] R. He, Z. Zhong, B. Ai, and J. Ding, “An empirical path loss model
IEEE Commun. Mag., vol. 45, no. 2, pp. 56–62, Feb. 2007. and fading analysis for high-speed railway viaduct scenarios,” IEEE
[40] J.-Y. Zhang, Z.-H. Tan, and X.-X. Yu, “Coverage efficiency of radio- Antennas Wireless Propag. Lett., vol. 10, pp. 808–812, Aug. 2011.
over-fiber network for high-speed railways,” in Proc. IEEE 6th Int. [63] H. Wei, Z. Zhong, L. Xiong, B. Ai, and R. He, “Study on the shadow
Conf. Wireless Commun. Netw. Mobile Comput. (WiCOM’10), Chengdu, fading characteristic in viaduct scenario of the high-speed railway,” in
China, Sep. 2010, pp. 1–4. Proc. CHINACOM, Harbin, China, Aug. 2011, pp. 1216–1220.
[41] ITU-R M.2135-1, “Guidelines for evaluation of radio interface technolo- [64] B. Zhang, Z. Zhong, B. Ai, D. Yao, and R. He, “Measurements and
gies for IMT-advanced,” Geneva, Switzerland, Rep. ITU-R M.2135-1, modeling of cross-correlation property of shadow fading in high-speed
Dec. 2009. railways,” in Proc. IEEE Veh. Technol. Conf. (VTC’14-Fall), Vancouver,
[42] P. Kyösti et al., “WINNER II channel models,” IST-4-027756, BC, Canada, Sep. 2014, pp. 1–5.
WINNER II D1.1.2, v1.2, Apr. 2008. [65] L. Gao, Z. Zhong, B. Ai, L. Xiong, and H. Zhang, “Analysis and
[43] B. Ai et al., “Radio wave propagation scene partitioning for high- emulation of the small-scale fading characteristics in the high-speed
speed rails,” Int. J. Antennas Propag., vol. 2012, 7 p., 2012, doi: rail scenarios,” in Proc. CHINACOM, Harbin, China, Aug. 2011,
10.1155/2012/815232. pp. 1181–1185.
[44] 3GPP, “Universal mobile telecommunications system (UMTS); Radio [66] L. Liu et al., “Position-based modeling for wireless channel on high-
frequency (RF) system scenarios (Release 12),” TR 25.942, version speed railway under a viaduct at 2.35 GHz,” IEEE J. Sel. Areas
12.0.0, Oct. 2014. Commun., vol. 30, no. 4, pp. 834–845, May 2012.
[45] L. Liu et al., “The COST 2100 MIMO channel model,” IEEE Wireless [67] Y. Guo, J. Zhang, C. Zhang, and L. Tian, “Correlation analysis of high-
Commun., vol. 19, no. 6, pp. 92–99, Dec. 2012. speed railway channel parameters based on channel measurement,” in
[46] “Requirements for the radio interface(s) for international mobile Proc. Int. Workshop High Mobility Wireless Commun. (HMWC’13),
telecommunications-2000 (IMT-2000),” ITU, Geneva, Switzerland, Shanghai, China, Nov. 2013, pp. 1–5.
Tech. Rep. ITU-R M.1034-1, 2000. [68] R. He, Z. Zhong, B. Ai, and J. Ding, “Measurements and analysis
[47] R. He, Z. Zhong, B. Ai, and K. Guan, “Reducing the cost of high- of short-term fading behavior for high-speed rail viaduct scenario,” in
speed railway communications: From the propagation channel view,” Proc. IEEE Int. Conf. Commun. (ICC’12), Ottawa, ON, Canada, Jun.
IEEE Trans. Intell. Transp. Syst. vol. 16, no. 4, pp. 2050–2060, Aug. 2012, pp. 4563–4567.
2015. [69] R. He, Z. Zhong, B. Ai, G. Wang, J. Ding, and A. F. Molisch,
[48] J. Rodríguez-Piñeiro, J. A. García-Naya, Á. Carro-Lagoa, and “Measurements and analysis of propagation channels in high-speed
L. Castedo, “A testbed for evaluating LTE in high-speed trains,” in Proc. railway viaducts,” IEEE Trans. Wireless Commun., vol. 12, no. 2,
IEEE Euromicro Conf. Digit. Syst. Des. (DSD’13), Santander, Spain, pp. 794–805, Feb. 2013.
Sep. 2013, pp. 175–182. [70] Q. Wang, C. Xu, M. Zhao, and D. Yu “Results and analysis for a
[49] N. Kita et al., “Experimental study of propagation characteristics for novel 2x2 channel measurement applied in LTE-R at 2.6 GHz,” in Proc.
wireless communications in high-speed train cars,” in Proc. 3rd Eur. IEEE Wireless Commun. Netw. Conf. (WCNC’14), Istanbul, Turkey,
Conf. Antennas Propag. (EuCAP’09), Berlin, Germany, Mar. 2009, Apr. 2014, pp. 177–181.
pp. 897–901. [71] R. Sun, C. Tao, L. Liu, Z. Tan, L. Zhang, and T. Zhou, “Nonisotropic
[50] W. Dong, G. Liu, L. Yu, H. Ding, and J. Zhang, “Channel prop- scattering characteristic in an alternant tree-blocked viaduct scenario on
erties of indoor part for high-speed train based on wideband chan- high-speed railway at 2.35 GHz,” Int. J. Antennas Propag., vol. 2014, 9
nel measurement,” in Proc. CHINACOM, Beijing, China, Aug. 2010, p., 2014, doi: 10.1155/2014/642894.
pp. 1–4. [72] T. Zhou, C. Tao, L. Liu, and Z. Tan, “A semiempirical MIMO channel
[51] M. Zhao et al., “Analysis and modeling for train-ground wireless model in obstructed viaduct scenarios on high-speed railway,” Int. J.
wideband channel of LTE on high-speed railway,” in Proc. IEEE Antennas Propag., vol. 2014, 10 p., 2014, doi: 10.1155/2014/287159.
Veh. Technol. Conf. (VTC’13-Spring), Dresden, Germany, Jun. 2013, [73] J. Lu, G. Zhu, and C. Briso-Rodriguez, “Fading characteristics in the
pp. 1–5. railway terrain cuttings,” in Proc. IEEE Veh. Technol. Conf. (VTC’11-
[52] R. Parviainen, P. Kyosti, Y. Hsieh, P. Ting, and J. Chiou, “Results of Spring), Budapest, Hungary, May 2011, pp. 1–5.
high speed train channel measurements,” in Proc. COST (TD’08), Lille, [74] R. He, Z. Zhong, B. Ai, and J. Ding, “Propagation measurements
France, Oct. 2008. and analysis for high-speed railway cutting scenario,” Electron. Lett.,
[53] K. Pekka, “WINNER II channel models—Part II radio channel mea- vol. 47, no. 21, pp. 1167–1168, Oct. 2011.
surement and analysis results,” IST-4-027756, WINNER II D1.1.2, v1.0, [75] R. He, Z. Zhong, B. Ai, J. Ding, and Y. Yang, “Propagation mea-
Sep. 2007. surements and analysis of fading behavior for high speed rail cutting
[54] R. He et al., “Measurement based channel modeling with directional scenarios,” in Proc. IEEE GLOBALCOM, Anaheim, CA, USA, Dec.
antennas for high-speed railways,” in Proc. IEEE Wireless Commun. 2012, pp. 5015–5020.
Netw. Conf. (WCNC’13), Shanghai, China, Apr. 2013, pp. 2932–2936. [76] R. He, Z. Zhong, B. Ai, J. Ding, Y. Yang, and A. F. Molisch, “Short-term
[55] H. Wei, Z. Zhong, K. Guan, and B. Ai, “Path loss models in viaduct fading behavior in high-speed railway cutting scenario: Measurements,
and plain scenarios of the high-speed railway,” in Proc. CHINACOM, analysis, and statistical models,” IEEE Trans. Antennas Propag., vol. 61,
Beijing, China, Aug. 2010, pp. 1–5. no. 4, pp. 2209–2222, Apr. 2013.
[56] J. Lu, G. Zhu, and B. Ai, “Radio propagation measurements and mod- [77] R. Sun, C. Tao, L. Liu, and Z. Tan, “Channel measurement and charac-
eling in railway viaduct area,” in Proc. IEEE 6th Int. Conf. Wireless terization for HSR U-shape groove scenarios at 2.35 GHz,” in Proc.
Commun. Netw. Mobile Comput. (WiCOM’10), Chengdu, China, Sep. IEEE Veh. Technol. Conf. (VTC’13-Fall), Las Vegas, NV, USA, Sep.
2010, pp. 1–5. 2013, pp. 1–5.
[57] L. Gao, Z. Zhong, B. Ai, and L. Xiong, “Estimation of the Ricean factor [78] L. Tian, J. Zhang, and C. Pan, “Small scale fading characteristics of
in K the high speed railway scenarios,” in Proc. CHINACOM, Beijing, wideband radio channel in the U-shape cutting of high-speed railway,”
China, Aug. 2010, pp. 1–5. in Proc. IEEE Veh. Technol. Conf. (VTC’13-Fall), Las Vegas, NV, USA,
[58] R. He, Z. Zhong, B. Ai, and J. Ding, “Distance-dependent model of Sep. 2013, pp. 1–6.
Ricean K-factors in high-speed rail viaduct channel,” in Proc. IEEE Veh. [79] F. Luan, Y. Zhang, L. Xiao, C. Zhou, and S. Zhou, “Fading char-
Technol. Conf. (VTC’12-Fall), Québec City, QC, Canada, Sep. 2012, acteristics of wireless channel on high-speed railway in hilly ter-
pp. 1–5. rain scenario,” Int. J. Antennas Propag., vol. 2013, 9 p., 2013, doi:
[59] R. He, Z. Zhong, and B. Ai, “Path loss measurements and analy- 10.1155/2013/378407.
sis for high-speed railway viaduct scene,” in Proc. 6th Int. Wireless [80] Y. Zhang, Z. He, W. Zhang, L. Xiao, and S. Zhou, “Measurement-
Commun. Mobile Comput. Conf. (IWCMC’10), Caen, France, Jul. 2010, based delay and Doppler characterizations for high-speed railway
pp. 266–270. hilly scenario,” Int. J. Antennas Propag., vol. 2014, 8 p., 2014, doi:
[60] R. He, Z. Zhong, B. Ai, L. Xiong, and H. Wei, “A novel path loss model 10.1155/2014/875345.
for high-speed railway viaduct scenarios,” in Proc. IEEE 7th Int. Conf. [81] P. Aikio, R. Gruber, and P. Vainikainen, “Wideband radio channel
Wireless Commun. Netw. Mobile Comput. (WiCOM’11), Wuhan, China, measurements for train tunnels,” in Proc. IEEE Veh. Technol. Conf.
Sep. 2011, pp. 1–4. (VTC’98), Ottawa, ON, Canada, May 1998, pp. 460–464.
[82] C. Briso-Rodriguez, J. M. Cruz, and J. I. Alonso, “Measurements and [104] K. Guan, Z. Zhong, B. Ai, and T. Kürner, “Deterministic propagation
modeling of distributed antenna systems in railway tunnels,” IEEE modeling for the realistic high-speed railway environment,” in Proc.
Trans. Veh. Technol., vol. 56, no. 5, pp. 2870–2879, Sep. 2007. IEEE Veh. Technol. Conf. (VTC’13-Spring), Dresden, Germany, Jun.
[83] K. Guan, Z. Zhong, B. Ai, and T. Kürner, “Propagation measurements 2013, pp. 1–5.
and analysis for train stations of high-speed railway at 930 MHz,” IEEE [105] D. J. Cichon, T. C. Becker, and W. Wiesbeck, “Determination of
Trans. Veh. Technol., vol. 63, no. 8, pp. 3349–3516, Oct. 2014. time-variant radio links in high-speed train tunnels by ray optical mod-
[84] K. Guan, Z. Zhong, B. Ai, and T. Kurner, “Empirical models for extra eling,” in Proc. IEEE Antennas Propag. Soc. Int. Symp. (APS/URSI’95),
propagation loss of train stations on high-speed railway,” IEEE Trans. Newport Beach, CA, USA, Jun. 1995, pp. 508–511.
Antennas Propag., vol. 62, no. 3, pp. 1395–1408, Mar. 2014. [106] D. J. Cichon, T. Zwick, and W. Wiesbeck, “Ray optical modeling of
[85] J. Lu, G. Zhu, and B. Ai, “Fading analysis for the high speed rail- wireless communications in high-speed railway tunnels,” in Proc. IEEE
way viaduct and terrain cutting scenarios,” Int. J. Antennas Propag., Veh. Technol. Conf. (VTC’96-Spring), Atlanta, GA, USA, May 1996,
vol. 2012, 9 p., 2012. pp. 546–550.
[86] K. Guan, Z. Zhong, B. Ai, and T. Kürner, “Propagation measurements [107] B. Chen and Z. Zhong, “Geometry-based stochastic modeling for
and modeling of crossing bridges on high-speed railway at 930 MHz,” MIMO channel in high-speed mobile scenario,” Int. J. Antennas
IEEE Trans. Veh. Technol., vol. 63, no. 8, pp. 502–517, Feb. 2014. Propag., vol. 2012, 6 p., 2012, doi: 10.1155/2012/184682.
[87] Y. Wen, Y. Ma, X. Zhang, X. Jin, and F. Wang, “Channel fading statistics [108] Q. Zheng, C. Xu, and M. Wu, “A novel MIMO channel model for
in high-speed mobile environment,” in Proc. IEEE APS Topical Conf. high speed railway system,” in Proc. IEEE Int. Conf. Commun. Technol.
Antennas Propag. Wireless Commun. (APWC’12), Cape Town, South (ICCT’12), Chengdu, China, Nov. 2012, pp. 31–35.
Africa, Sep. 2012, pp. 1209–1212. [109] A. Ghazal, C.-X. Wang, H. Haas, M. A. Beach, X. Lu, and D. Yuan,
[88] B. Chen, Z. Zhong, and B. Ai, “Stationarity intervals of time-variant “A non-stationary MIMO channel model for high speed train commu-
channel in high speed railway scenario,” J. China Commun., vol. 9, nication systems,” in Proc. IEEE Veh. Technol. Conf. (VTC’12-Spring),
no. 8, pp. 64–70, Aug. 2012. Yokohama, Japan, May 2012, pp. 1–5.
[89] R. He, D. Zhong, B. Ai, and C. Oestges, “Shadow fading correlation in [110] A. Ghazal et al., “A non-stationary geometry-based stochastic model
high-speed railway environments,” IEEE Trans. Veh. Technol., vol. 64, for MIMO high-speed train channels,” in Proc. 12th Int. Conf. ITS
no. 7, pp. 2762–2772, Jul. 2015. Telecommun. (ITST’12), Taipei, Taiwan, Nov. 2012, pp. 7–11.
[90] R. He, Z. Zhong, B. Ai, and B. Zhang, “Measurement-based auto- [111] A. Ghazal, C.-X. Wang, B. Ai, D. Yuan, and H. Haas, “A non-stationary
correlation model of shadow fading for the high-speed railways in urban, wideband MIMO channel model for high-mobility intelligent trans-
suburban, and rural environments,” in Proc. Antennas Propag. Soc. Int. portation systems,” IEEE Trans. Intell. Transp. Syst., vol. 16, no. 2,
Symp. (APSURSI’14), Memphis, TN, USA, Jul. 2014, pp. 949–950. pp. 885–897, Apr. 2015.
[91] J. Qiu, C. Tao, L. Liu, and Z. Tan, “Broadband channel measurement for [112] A. Ghazal, C.-X. Wang, Y. Liu, P. Fan, and M. K. Chahine, “A generic
the high-speed railway based on WCDMA,” in Proc. IEEE Veh. Technol. non-stationary MIMO channel model for different high-speed train sce-
Conf. (VTC’12-Spring), Yokohama, Japan, May 2012, pp. 1–5. narios,” in Proc. IEEE Int. Conf. Commun. China (ICCC’15), Shenzhen,
[92] L. Liu, C. Tao, T. Zhou, Y. Zhao, X. Yin, and H. Chen, “A highly effi- China, Nov. 2015.
cient channel sounding method based on cellular communications for [113] B. Chen, Z. Zhong, B. Ai, and D. Michelson, “A geometry-based
high-speed railway scenarios,” EURASIP J. Wireless Commun. Netw., stochastic channel model for high-speed railway cutting scenarios,”
vol. 2012, 16 p., 2012, doi: 10.1186/1687-1499-2012-307. IEEE Antennas Wireless Propag. Lett., vol. 14, pp. 851–854, Apr.
[93] X. Yin, X. Cai, X. Cheng, J. Chen, and M. Tian, “Empirical geometry- 2015.
based random-cluster model for high-speed-train channels in UMTS [114] S. Lin, Z. Zhong, L. Cai, and Y. Luo, “Finite state Markov modelling
networks,” IEEE Trans. Intell. Transp. Syst., vol. 16, no. 5, pp. 2850– for high speed railway wireless communication channel,” in Proc. IEEE
2861, Oct. 2015. GLOBECOM, Anaheim, CA, USA, Dec. 2012, pp. 5421–5426.
[94] M. Dohler and Y. Li, Cooperative Communications: Hardware, Channel [115] S. Lin et al., “Finite-state Markov modeling for high-speed railway
& PHY. Hoboken, NJ, USA: Wiley, 2010. fading channels,” IEEE Antennas Wireless Propag. Lett., vol. 14,
[95] D. Wang, L. Song, X. Kong, and Z. Zhang, “Near-ground path loss mea- pp. 954–957, Apr. 2015.
surements and modeling for wireless sensor networks at 2.4 GHz,” Int. J. [116] L. Liu, C. Tao, J. Qiu, T. Zhou, R. Sun, and H. Chen, “The dynamic
Distrib. Sensor Net., vol. 2012, 10 p., 2012, doi: 10.1155/2012/969712. evolution of multipath components in high-speed railway in viaduct sce-
[96] K. Guan, Z. D. Zhong, J. I. Alonso, and C. Briso-Rodriguez, narios: From the birth-death process point of view,” in Proc. IEEE 23rd
“Measurement of distributed antenna system at 2.4 GHz in a realistic Int. Symp. Pers. Indoor Mobile Radio Commun. (PIMRC’12), Sydney,
subway tunnel environment,” IEEE Trans. Veh. Technol., vol. 61, no. 2, NSW, Australia, Sep. 2012, pp. 1774–1778.
pp. 834–837, Feb. 2012. [117] Y. Liu, C.-X. Wang, A. Ghazal, S. Wu, and W. Zhang, “A multi-
[97] K. Guan, Z. D. Zhong, and B. Ai, “Statistic modeling for propagation in mode waveguide tunnel channel model for high-speed train wireless
tunnels based on distributed antenna systems,” in Proc. IEEE Antennas communication systems,” in Proc. 9th Eur. Conf. Antennas Propag.
Propag. Soc. Int. Symp. (APSURSI’13), Orlando, FL, USA, Jul. 2013, (EuCAP’15), Lisbon, Portugal, Apr. 2015, pp. 1–5.
pp. 1920–1921. [118] R. He, Z. Zhong, and C. Briso, “Broadband channel long delay clus-
[98] W. Qian, X. Chunxiu, W. Muqing, Z. Min, and Y. Deshui, “Propagation ter measurements and analysis at 2.4 GHz in subway tunnels,” in Proc.
characteristics of high speed railway radio channel based on broadband IEEE Veh. Technol. Conf. (VTC’11-Spring), Budapest, Hungary, May
measurements at 2.6 GHz,” in Proc. IEEE Wireless Commun. Netw. 2011, pp. 1–5.
Conf. (WCNC’14), Istanbul, Turkey, Apr. 2014, pp. 166–170. [119] K. Guan, Z. Zhong, B. Ai, and T. Kurner, “Semi-deterministic path-loss
[99] T. Zhou, C. Tao, L. Liu, and Z. Tan, “Ricean K-factor measurements modeling for viaduct and cutting scenarios of high-speed railway,” IEEE
and analysis for wideband radio channels in high-speed railway U-shape Antennas Wireless Propag. Lett., vol. 12, pp. 789–792, Jun. 2013.
cutting scenarios,” in Proc. IEEE Veh. Technol. Conf. (VTC’14-Spring), [120] C.-X. Wang, M. Ptzold, and Q. Yao, “Stochastic modeling and
Seoul, Korea, May 2014, pp. 1–5. simulation of frequency correlated wideband fading channels,”
[100] L. Tian and J. Zhang, “Analysis on non-stationary characteristics of IEEE Trans. Veh. Technol., vol. 56, no. 3, pp. 1050–1063, May
wideband radio channel in HSR U-shape cutting scenario,” in Proc. 2007.
IEEE Int. Conf. Commun. China (ICCC’14), Shanghai, China, Oct. [121] M. Patzold, Mobile Radio Channels, 2nd ed. Hoboken, NJ, USA: Wiley,
2014, pp. 544–548. 2011.
[101] J. Yang, B. Ai, and Z. Zhong “Construction and capacity analysis of [122] R. He, B. Ai, Z. Zhong, A. F. Molisch, R. Chen, and Y. Yang,
high-rank LoS MIMO channels in high speed railway scenarios,” Int. J. “A measurement-based stochastic model for high-speed railway chan-
Antennas Propag., vol. 2012, pp. 1–4, 2012, doi: 10.1155/2012/423759. nels,” IEEE Trans. Intell. Transp. Syst., vol. 16, no. 3, pp. 1120–1135,
[102] L. Tian, X. Yin, Q. Zuo, J. Zhou, Z. Zhong, and S. Lu, “Channel Jun. 2015.
modeling based on random propagation graphs for high speed rail- [123] L. Xuan, S. Chao, B. Ai, and Z. Gang, “Finite-state Markov modeling of
way scenarios,” in Proc. IEEE 23rd Int. Symp. Pers. Indoor Mobile fading channels: A field measurement in high-speed railways,” in Proc.
Radio Commun. (PIMRC’12), Sydney, NSW, Australia, Sep. 2012, IEEE Int. Conf. Commun. China (ICCC’13), Xi’an, China, Aug. 2013,
pp. 1746–1750. pp. 577–582.
[103] S. Knörzer, M. A. Baldauf, T. Fugen, and W. Wiesbeck, “Channel anal- [124] 3GPP TR 36.873 V12.2.0, “Study on 3D channel model for LTE,” Jun.
ysis for an OFDM-MISO train communications system using different 2015.
antennas,” in Proc. IEEE Veh. Technol. Conf. (VTC’07-Fall), Baltimore, [125] B. Mondal et al., “3D channel model in 3GPP,” IEEE Commun. Mag.,
MD, USA, Oct. 2007, pp. 809–813. vol. 53, no. 3, pp. 16–23, Mar. 2015.
[126] M. Shafi, M. Zhang, P. J. Smith, A. L. Moustakas, and A. F. Molisch, for the IEEE J OURNAL ON S ELECTED A REAS IN C OMMUNICATIONS (spe-
“The impact of elevation angle on MIMO capacity,” in Proc. IEEE Int. cial issue on Vehicular Communications and Networks). He is a Fellow of the
Conf. Coomun. (ICC’06), Istanbul, Turkey, Jun. 2006, pp. 4155–4160. IET and the HEA, and a member of EPSRC Peer Review College. He served
[127] J. Meinila et al., “D5.3: WINNER+ final channel models,” or is serving as a TPC member, the TPC Chair, and the General Chair for over
CELTIC/CP5-026, Jun. 2010. 80 international conferences. He was the recipient of Best Paper Awards from
[128] T. A. Thomas, F. W. Vook, E. Mellios, G. S. Hilton, A. R. Nix, and the IEEE GLOBECOM 2010, the IEEE ICCT 2011, the ITST 2012, the IEEE
E. Visotsky, “3D extension of the 3GPP/ITU channel model,” in Proc. VTC 2013-Spring, and the IWCMC 2015.
IEEE Veh. Technol. Conf. (VTC’13-Spring), Dresden, Germany, Jun.
2013, pp. 1–5.
[129] R. Almesaeed, A. S. Ameen, E. Mellios, A. Doufexi, and A. R. Nix, “A Ammar Ghazal (S’15) received the B.Sc. degree
proposed 3D extension to the 3GPP/ITU channel model for 800 MHz in electronics and telecommunication engineering
and 2.6 GHz bands,” in Proc. 8th Eur. Conf. Antennas Propag. from Damascus University, Damascus, Syria, and
(EuCAP’14), The Hague, The Netherlands, Apr. 2014, pp. 3039–3043. the M.Sc. and Ph.D. degrees from Heriot-Watt
[130] P. Liu, B. Ai, Z. Zhong, and X. Zhou, “A novel train-to-train commu- University, Edinburgh, U.K., in 2006, 2010, and
nication model design based on multihop in high-speed railway,” Int. J. 2015, respectively. His research interests include
Antennas Propag., vol. 2012, 9 p., 2012, doi: 10.1155/2012/475492. wireless propagation channel characterization and
[131] P. Liu, X. Zhou, and Z. Zhong, “Outage analysis of train-to- modeling, nonstationary channel models, and high-
train communication model over Nakagami-m channel in high-speed speed train wireless propagation.
railway,” Int. J. Antennas Propag., vol. 2013, 10 p., 2013, doi:
10.1155/2013/617895.
[132] Q. Luo, W. Fang, T. Yang, and D. Wang, “A transmit beamforming algo-
rithm for high-speed train communication,” in Proc. IEEE Veh. Technol.
Conf. (VTC’12-Fall), Québec City, QC, Canada, Sep. 2012, pp. 1–5. Bo Ai (M’00–SM’09) received the master’s and
[133] H.-H. Wang and H.-A. Hou, “Experimental analysis of beamform- Ph.D. degrees from Xidian University, Xi’an, China,
ing in high-speed railway communication,” in Proc. IEEE 22nd Int. in 2002 and 2004, respectively. He is currently work-
Symp. Pers. Indoor Mobile Radio Commun. (PIMRC’11), Toronto, ON, ing with Beijing Jiaotong University, Beijing, China,
Canada, Sep. 2011, pp. 745–749. as a Professor and an Advisor of Ph.D. candidates.
[134] H. Zhu, “Radio resource allocation for OFDMA systems in high speed He is the Deputy Director of State Key Laboratory
environments,” IEEE J. Sel. Areas Commun., vol. 30, no. 4, pp. 748– of Rail Traffic Control and Safety, Beijing Jiaotong
759, May 2012. University. He has authored/coauthored five books,
[135] M. Cheng, X. Fang, and L. Yan, “Beamforming and Alamouti STBC 26 invention patents, and 220 scientific research
combined downlink transmission schemes in communication systems papers in his research area till now. His research inter-
for high-speed railway,” in Proc. Int. Conf. Wireless Commun. Signal ests include broadband and dedicated mobile commu-
Process. (WCSP’13), Hangzhou, China, Oct. 2013, pp. 1–6. nication theories and technologies for GSM-R and LTE-R, with emphasis on
[136] J. Yao, S. S. Kanhere, and M. Hassan, “Mobile broadband performance channel measurements and modeling. He received the Excellent Postdoctoral
measured from high-speed regional trains,” in Proc. IEEE Veh. Technol. Research Fellow (with great Hons.) from Tsinghua University, Beijing, China,
Conf. (VTC’11-Fall), San Francisco, CA, USA, Sep. 2011, pp. 1–5. in 2007. He is a Fellow of the IET. He is an Associate Editor for the IEEE
[137] Y. Dong, P. Fan, and K. Letaief, “High speed railway wireless commu- T RANSACTIONS ON C ONSUMER E LECTRONICS and an Editorial Committee
nications: Efficiency v.s. fairness,” IEEE Trans. Veh. Technol., vol. 63, Member of the journal, Wireless Personal Communications.
no. 2, pp. 925–930, Feb. 2014.
[138] B. Chen, Z. Zhong, B. Ai, and X. Chen, “Comparison of antenna arrays
for MIMO system in high speed mobile scenarios,” in Proc. IEEE
Veh. Technol. Conf. (VTC’11-Spring), Budapest, Hungary, May 2011,
pp. 1–5. Yu Liu received the B.Sc. and M.Eng. degrees
[139] Y. Fu, C.-X. Wang, A. Ghazal, H. Aggoune, and M. M. Alwakeel, in communication and information systems from
“Performance investigation of spatial modulation systems under non- Qufu Normal University, Qufu, China, in 2010 and
stationary wideband high-speed train channel models,” IEEE Trans. 2013, respectively. She is currently pursuing the
Wireless Commun., submitted for publication. Ph.D. degree at the School of Information Science
and Engineering (ISE), Shandong University, Jinan,
China. Her research interests include nonstationary
channel modeling, high-speed train wireless propa-
Cheng-Xiang Wang (S’01–M’05–SM’08) received gation characterization and modeling, and channel
the B.Sc. and M.Eng. degrees in communication modeling for special scenarios.
and information systems from Shandong University,
Jinan, China, and the Ph.D. degree in wireless
communications from Aalborg University, Aalborg,
Denmark, in 1997, 2000, and 2004, respectively. He Pingzhi Fan (M’93–SM’99–F’15) received the
has been with Heriot-Watt University, Edinburgh, Ph.D. degree in electronic engineering from Hull
U.K., since 2005, and was promoted to a Professor University, Kingston upon Hull, U.K. He is cur-
of wireless communications in 2011. He is also an rently a Professor and the Director of the Institute
Honorary Fellow with the University of Edinburgh, of Mobile Communications, Southwest Jiaotong
Edinburgh, U.K., and a Chair/Guest Professor with University, Chengdu, China. He was the Chief
Shandong University and Southeast University, Nanjing, China. He was a Scientist of a National 973 Research Project. He has
Research Fellow with the University of Agder, Grimstad, Norway, from 2001 to over 200 research papers published in various aca-
2005, a Visiting Researcher at Siemens AG-Mobile Phones, Munich, Germany, demic English journals (IEEE/IEE/IEICE, etc.), and
in 2004, and a Research Assistant with the Technical University of Hamburg- eight books (including edited), and is the inventor of
Harburg, Hamburg, Germany, from 2000 to 2001. He has edited one book and 20 granted patents. His research interests include high
published over 230 papers in refereed journals and conference proceedings. His mobility wireless communications, 5G technologies, wireless networks for big
research interests include wireless channel modelling and 5G wireless commu- data, and signal design and coding. He served as the General Chair or TPC
nication networks, including green communications, cognitive radio networks, chair of a number of international conferences, and has been the Guest Editor-
high mobility communication networks, massive MIMO, millimeter wave com- in-Chief, the Guest Editor or the Editorial Member of several international
munications, and visible light communications. He served or is currently serv- journals. He is the Founding Chair of the IEEE VTS BJ Chapter and ComSoc
ing as an Editor for international journals, including the IEEE T RANSACTIONS CD Chapter, the Founding Chair of the IEEE Chengdu Section. He also served
ON V EHICULAR T ECHNOLOGY (since 2011), the IEEE T RANSACTIONS as a Board Member of the IEEE Region 10, the IET (IEE) Council and the
ON C OMMUNICATIONS (since 2015), and the IEEE T RANSACTIONS ON IET Asia-Pacific Region. He was a recipient of the U.K. ORS Award and the
W IRELESS C OMMUNICATIONS (2007–2009). He was the leading Guest Editor Outstanding Young Scientist Award by NSFC.

Channel Measurements and Models For High-Speed Train Communication Systems A Survey

Uploaded by

Copyright:

Available Formats

Channel Measurements and Models For High-Speed Train Communication Systems A Survey

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Channel Measurements and Models For High-Speed Train Communication Systems A Survey

Uploaded by

Copyright:

Available Formats

974 IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 18, NO.

2, SECOND QUARTER 2016

Channel Measurements and Models for High-Speed

networks scenario in the WINNER II channel model [42] and

A. HST Propagation Scenarios

Fig. 3. Classification of non-stationary HST channel models.

C. HST Scenarios obstacles and accidents via wireless communication links. A

You might also like