Content uploaded by Ke Guan
Author content
All content in this area was uploaded by Ke Guan on Dec 14, 2015
Content may be subject to copyright.
IEEE Communications Magazine • October 2015
78 0163-6804/15/$25.00 © 2015 IEEE
Bo Ai, Ke Guan, Qi
Wang, Guo-Yu Ma, Yan
Li, Lei Xiong and Jian-
Wen Ding are with Beijing
Jiaotong University.
Bo Ai (corresponding
author) is also a visiting
professor with the Depart-
ment of Electrical Engi-
neering, Stanford
University.
Markus Rupp is with TU
Wien.
Thomas Kürner is with
Technical University
Braunschweig.
Xiang Cheng is with
Peking University.
Xue-Feng Yin is with
TongJi University.
ABSTRACT
The future development of the railway is
highly desired to evolve into a new era where
infrastructure, trains, travelers, and goods will be
increasingly interconnected to provide high com-
fort, with optimized door-to-door mobility at
higher safety. For this vision, it is required to
realize seamless high data rate wireless connec-
tivity for railways. To improve the safety and
comfort of future railways, wireless communica-
tions for railways are required to evolve from
only voice and traditional train control signaling
services to various high data rate services includ-
ing critical high-definition (HD) video and other
more bandwidth-intensive passenger services,
such as onboard and wayside HD video surveil-
lance, onboard real-time high data rate services,
train multimedia dispatching video streaming,
railway mobile ticketing, and the Internet of
Things for railways. Corresponding mobile com-
munications network architecture under various
railway scenarios including inter-car, intra-car,
inside station, train-to-infrastructure and infra-
structure-to-infrastructure are proposed in this
article. Wireless coverage based on massive
MIMO for railway stations and train cars is pro-
posed to fulfill the requirement of high-data-rate
and high spectrum efficiency. The technical chal-
lenges brought by the massive MIMO technique
are discussed as well.
INTRODUCTION
High-speed railway (HSR), intercity railway, sub-
way, light rail, and other rail traffic systems have
brought much convenience for people’s travel
with less energy consumption and air pollution
compared to cars and airplanes. Moreover, HSR
moves very fast with high comfort and high
punctuality. To ensure safe and reliable opera-
tion of railways, the train operation control sys-
tem acts as a nerve center. To make such a nerve
center work well, a reliable bidirectional commu-
nication link between the train and the ground is
of great importance. Global System for Mobile
Communications for Railway (GSM-R) plays a
key role in realizing such bidirectional communi-
cations. Train timetable information, the driving
license, train speed, train location, and other
train control signals can be transmitted through
a GSM-R network.
With the increasing demand for new railway
services such as railway multimedia dispatching
communication and railway emergency commu-
nication, Long Term Evolution for Railway
(LTE-R) is now under discussion [1, 2]. Such a
broadband communication system has the capa-
bility of 100 MHz data transmission rate in high
mobility with 20 MHz bandwidth. However, just
as those topics specified in the European
“Shift2Rail” project (http://www.shift2rail.org),
intelligent rail infrastructure, intelligent mobility
management, smart rail services, and a new gen-
eration of rail vehicles ultimately form the
requirements of a seamless high data rate wire-
less connectivity for future rail development.
Thus, higher-frequency-band techniques such as
millimeter-wave (mmWave), the fifth generation
(5G), and the corresponding mobile communica-
tion network should be designed accordingly to
provide high capacity and high data rate for
future railway services. As far as the authors
know, although plenty of literature has discussed
GSM-R communication networks [3] and 5G
techniques [4], no literature deals with the com-
munications network regarding future railway
services.
The new services for railways may pose spe-
cial requirements for the new mobile communi-
cations network architecture. Future railway
services and typical communications scenarios
are described. The heterogeneous mobile net-
work architecture for future railway system,
related promising key technologies, and techni-
cal challenges are discussed. Massive multiple-
input multiple-output (MIMO)-based wireless
coverage for railway stations, inside cars, and
other railway hotspot areas is discussed in detail.
Finally, conclusions are drawn.
FUTURE RAILWAY SERVICES AND
BANDWIDTH REQUIREMENT
With numerous HSRs being operated and
planned in China and Europe in particular, reli-
able communications for railway control and
safety are of great current and future interest. In
addition, high efficiency, environment friendli-
FUTURE RAILWAY COMMUNICATIONS
Bo Ai, Ke Guan, Markus Rupp, Thomas Kürner, Xiang Cheng, Xue-Feng Yin, Qi Wang, Guo-Yu Ma, Yan Li,
Lei Xiong, and Jian-Wen Ding
Future Railway Services-Oriented
Mobile Communications Network
AI_LAYOUT.qxp_Author Layout 10/1/15 3:38 PM Page 78
IEEE Communications Magazine • October 2015 79
ness, and passenger convenience are also goals
of future railway developments. In 2010, the E-
train project (http://www.uic.org/etf/publication/
publication-detail.php?code_pub=190_14) from
the International Union of Railways (UIC) sum-
marized over 200 railway services including train
dispatching, train control, train operation com-
munication, train state monitoring, and so on. In
2014, Horizon 2020 [5] emphasized that there
should be more railway services focused on real-
izing the objective of “smart, green and integrat-
ed transport.” To sum up, even though the
LTE-R network is designed to bear many rail-
way services, real-time HD video transmissions
supporting automatic driving, security closed cir-
cuit television (CCTV) in the train, remote
maintenance of trains, and other high data rate
railway services still challenge the current mobile
communications network for railways.
SERVICES FOR THE FUTURE RAILWAY
In order to realize the above-mentioned vision,
communications for future railways are required
to evolve from only critical signaling applications
to various high data rate services: onboard and
wayside HD video surveillance, onboard real-
time high data rate services, train multimedia
dispatching video streams, railway mobile ticket-
ing, and the Internet of Things for railways.
Onboard and wayside HD video surveillance:
Video surveillance services capture live 720p or
1080p true HD video images from high-defini-
tion television (HDTV) IP cameras and high-
definition serial digital interface (HD-SDI)
cameras located on trains (for train operation)
and the wayside along rail tracks (for asset con-
dition monitoring). In light of security concerns
(terrorist attacks, riots, and emergencies), the
HD video images are required to be both stored
locally and delivered in real time from trains/
wayside monitors to local and central train con-
trol centers (TCCs).
Onboard real-time high data rate services:
One of the most attractive future railway ser-
vices is the wireless Internet inside train cars.
Not only surfing the wireless Internet, having a
pleasant chat through Facebook or Twitter social
networks [6], passengers onboard also expect to
access real-time HD video for business and
entertainment such as on-vehicle video confer-
ence and live broadcast.
Train multimedia dispatching video stream:
Comprehensive dispatching information includ-
ing written words, data, voices, and images is
provided by a multimedia dispatching command
communication system to the dispatcher. For
instance, regarding automatic driving, an inten-
sive train multimedia dispatching video stream
of doorways for driverless trains is required by
dispatchers from remote TCCs to ensure that
the train doorways are clear before a train sets
off.
Railway mobile ticketing: To ensure the secu-
rity and reliability of the passenger ticketing sys-
tem, a dedicated railway communication network
will be used for the ticketing system, and a hand-
held ticketing terminal, through which identity
identification, check-in, and ticket selling will be
realized. This will alleviate the crowdedness of
the ticket hall and ease passenger traveling.
Internet of Things for railways: In addition to
real-time query and tracking the whole process
of the location of the train and goods, the Inter-
net of Things for railways can be developed to
integrate the sensing information of rail infras-
tructures including bridges, viaducts, tunnels,
leaky feeders, rail gaps, frozen soil, and slope
protection through various sensing measures
such as infrared, sound sensors, and temperature
sensors. The information is collected and sent
back to the computing center. On-time forecast-
ing and management decisions can be made
through big data or cloud computing platforms
to ensure safe operation of the train.
BANDWIDTH REQUIREMENT OF THE
FUTURE RAILWAY
The determination of the bandwidth depends
mainly on the services and the number of users.
The services consuming most of the bandwidth
should be analyzed. For an analysis of the num-
ber of users, the rules of simultaneous communi-
cation users within certain scenarios and
different service types should be followed.
Taking the real-time HD video broadband
connections inside a car in future railway ser-
vices into consideration, data rates of up to 3
Gb/s (allowing 1920 ×1080 @ 60 Hz @ 24 bits)
in a 40 MHz channel can be supported by the
wireless home display interface (WHDI). With
around 130–180 passengers/seats (in recent high-
speed and intercity trains and some double-deck
trains), up to 3.6 GHz total bandwidth is
required for one car if 50 percent of the passen-
gers want to have real-time HD video service for
business or entertainment. If bidirectional HD
video streaming is expected (for video confer-
ence), the bandwidth requirement may be dou-
ble (up to 7.2 GHz required). Therefore, an
LTE-R system with 20 MHz bandwidth cannot
support such services, and we have to resort to
the mmWave frequency bands and the 5G com-
munication system for railway (5G-R), which
provide large bandwidth and high data rate
transmission capability. The METIS project
group regards high mobility scenarios as typical
of 5G. The 5G promotion group of IMT-2020 in
China released a white paper on 5G vision and
requirements, in which the railway and subway
are defined as two typical scenarios for 5G. The
subway represents the 5G typical character of
super high density with over 6 persons/m2, and
the high-speed railway represents the typical
character of high mobility with speeds up to 350
km/h or above.
In general, the requirement of future railway
services and typical communication scenarios
calls for large bandwidth and high data rate
transmission capabilities.
MOBILE COMMUNICATIONS
NETWORK ARCHITECTURE
To represent the above service space, five com-
munication scenarios can be defined: train-to-
infrastructure, inter-car, intra-car, inside station,
and infrastructure-to-infrastructure. Although
technologies for the train-to-infrastructure sce-
The determination of
the bandwidth
depends mainly on
the services and the
number of users.
The services consum-
ing most of the
bandwidth should be
analyzed. For an
analysis of the num-
ber of the users, the
rules of simultaneous
communication users
within certain scenar-
ios and different ser-
vice types should be
followed.
AI_LAYOUT.qxp_Author Layout 10/1/15 3:38 PM Page 79
IEEE Communications Magazine • October 2015
80
nario have been investigated, very few technolo-
gies have been examined for the inter-car and
intra-car scenarios. This makes it very difficult to
build a seamless network supporting all five
communication scenarios for railways with the
current technologies.
RAILWAY COMMUNICATION SCENARIOS
The five communication scenarios for future
railways are described in detail as follows.
Train-to-infrastructure: Two kinds of links
are required between the access points (APs)/
transceivers of the train and the infrastructures
of fixed networks. The links provide bidirection-
al streams with high data rates and low latencies,
as well as robust communication links with laten-
cies lower than 100 ms together with an avail-
ability of 98–99 percent, while moving at speeds
up to 350 km/h or above.
Inter-car: A wireless network runs between
cars to avoid the high expense of wiring a train
for network access and the inconvenience of
rewiring when a train is reconfigured. This sce-
nario requires a high data rate and low latency
because the APs are arranged in each car such
that each AP serves as a client station for the
AP in the previous car, while also serving as an
AP for all the stations within its car.
Intra-car: The links provide wireless access
between the APs in the car and the passengers
or sensors of equipment inside the car. In this
scenario, real-time HD videos need to be
accessed with low latencies.
Inside the station: The links provide wireless
access between the APs and the user equipment
in railway stations. Users are strongly interested
in access to mobile broadband communication
services. The stations provide a fixed/wireless
communication infrastructure to support general
commercial (e.g., cash desks) as well as opera-
tional services (e.g., automatic doors, surveil-
lance, fire protection).
Infrastructure-to-infrastructure: HD video
and other information is transmitted in real time
among multiple HDTV IP/HD-SDI cameras,
and the APs deployed on the trains, on station
platforms, and the wayside along rail tracks, as a
high date rate wireless backhaul or the Internet
of Things. Infrastructures are real-time connect-
ed and interactive, supported by bidirectional
data streams with very high data rate and low
latencies.
MOBILE COMMUNICATIONS
NETWORK ARCHITECTURE
Considering a variety of railway services and sce-
narios, the corresponding mobile communications
network architecture should be heterogeneous
including various types of access networks working
at different frequency bands, satisfying multiple
bands, multiple scenarios, and various require-
ments of wireless coverage. Apart from satellite
communications, which can provide medium- and
high-capacity services for railways and will become
more important in the future, the architecture of a
heterogeneous land mobile communication net-
work for future railway systems is shown in Fig. 2.
Such a network will be composed of a macro base
station used for macrocell coverage, and some
micro base stations as backhaul to offer hotspot
area coverage. Transferring from one network
mode to another requires much time to adapt
even with a self-organizing network (SON) [1].
Therefore, how to realize seamless switching
between different network modes with low latencies
is really a challenging task.
The dedicated mobile communication net-
works, including GSM-R, LTE-R, and 5G-R,
and the public mobile communication network
in such heterogeneous architecture are repre-
sented as and in Fig. 2, respectively. The
dedicated networks cover the railway lines, rail-
way station, freight station, marshaling station,
railway hub, and other railway areas for train
operation control purposes. The communication
performance should satisfy the reliability, avail-
ability, maintenance, and safety (RAMS) from a
UIC standard or specification. Moreover, HSR
operates at extremely high speeds, introducing
high Doppler spreads and frequent handovers.
Thus, how to plan the distance between two adja-
cent base stations efficiently is another challenging
task. It should guarantee both enough time for
handover under high mobility and good cell edge
coverage to avoid call drops.
Figure 1. Panorama of seamless high data rate wireless connectivity for railways.
2
33
•••
•••
1
11
1
1 Train-to-infrastructure
2 Inter-car
3 Intra-car
4 Inside station
5 Infrastructure-to-infrastructure
11
5
5
4Core network
(Internet)
2
Considering a variety
of railway services
and scenarios, the
corresponding
mobile communica-
tions network archi-
tecture should be
heterogeneous
including various
types of access net-
works working at
different frequency
bands, satisfying
multi-band, multi-
scenarios and various
requirements of
wireless coverage.
AI_LAYOUT.qxp_Author Layout 10/1/15 3:38 PM Page 80
IEEE Communications Magazine • October 2015 81
Adjacent base stations of a GSM-R network,
denoted by in Fig. 2, are 3–4 km apart to
guarantee good cell edge coverage. The traffic
data is transmitted to a base station controller
(BSC) from a mobile service switching center
(MSC) of the core network and finally to the
radio remote unit (RRU) from the building
baseband unit (BBU). The GSM-R system is
nearly mature. However, the channel models for
GSM-R under various railway scenarios are not
perfect. Moreover, GSM-R should be compatible
and interconnected with an LTE-R network. LTE-
R, denoted by , will probably adopt 450 or 800
MHz frequency bands in China. The distance
between two adjacent base stations may be 7–8
km for 450 MHz and 3–4 km for 800 MHz to
ensure both enough handover time and good
quality of cell edge coverage. LTE-R uses the
flat network structure based on IP. The traffic
data is transmitted to a BBU from the Evolved
Packet Core (EPC). EPC is the architecture for
the convergence of voice in an LTE network.
The BBU then transmits data to the RRU. The
unsolved issues for LTE-R are propagation char-
acteristics and channel models at 450 and 800
MHz, the key techniques adaptive to high mobility.
A 5G-R network, represented by in Fig. 2,
is developed to cater for large bandwidth and
high data rate services. Due to its characteristic
high frequency bands and reduced coverage
capability, 5G-R is not appropriate for railway
line coverage, but for railway hotspot areas such
as railway stations and intra-car. As for the
implementation of 5G-R, the best candidate
technique is massive MIMO, which is regarded
as one of the key techniques for 5G. However,
the majority of the current research work on
massive MIMO has focused on static channel
conditions. One of the challenging tasks for 5G-
R is propagation characteristic and channel
models for massive MIMO dynamic channels
under various railway scenarios at 6, 28, 38, 60,
and 300 GHz frequency bands.
The railway station scenario denoted by in
Fig. 2 can be equipped with massive MIMO to
provide high capacity and high data rate trans-
mission for passengers inside the station. There
can be one macro base station with several micro
base stations installed in the corners inside the
railway station, forming a wireless backhaul. The
key technique that remains to be solved is the
propagation characteristics and channel models for
massive-MIMO-based coverage under the railway
station scenario. The main duty of the public
mobile communications network is recreation
such as wireless Internet access for passengers.
A challenging task for the coexistence of public
and dedicated mobile communication networks
is how to avoid the serious adjacent channel inter-
ference. In the future, the railway, subway, and
other rail systems may be combined with road
traffic to form an integrated transportation net-
work, where not only vehicle-to-vehicle (V2V),
but also train-to-train (T2T), train-to-infra-
structure (T2I), and train-to-vehicle (T2V) com-
munications will be included. T2V
communication is useful when the train passes
through the intersection of a rail track and a
road. T2T communication, represented by ,
means the direct communication between two
trains without infrastructure and any other APs.
The most challenging tasks for integrated trans-
portation are channel characteristics and channel
models under the T2T scenario, the categorizing of
T2T transmitted messages, the media access con-
trol (MAC) layer routing mechanism, and the key
techniques for long-range communications with
high mobility. represents the ground-to-ground
communications of the infrastructures along the
rail tracks. The infrastructure along the rail
tracks refers to the access points or webcams.
MASSIVE-MIMO-BASED
WIRELESS COVERAGE AND
TECHNICAL CHALLENGES
In the railway station and inside the car, where
the transmission rate and system capacity need
to be improved, massive MIMO can be used,
which has been proved theoretically to achieve
high data rate, high spectral efficiency, and high
energy efficiency [7]. These gains come from its
diversity and beamforming. Transmission and
receiving modes can be adjusted to be adaptable
Figure 2. Heterogeneous mobile communication network for future railway systems.
XX
station
RRU RRU
RRURRURRURRURRU
LTE-R
RRURRU
••• •••
•••
•••
••••••
•••
RRU BBU BBU BBU
BBU
GSM-R
Transmission beam without
significant Doppler shift
BBU BBU
7-8km
3-4km
EPC
4
87
6
5
1
2
3
8
5G-R
89
10
BSC MSC
Train station
Interference beam
Transmission beam with
significant Doppler shift
One of the
challenging tasks
for 5G-R are
propagation
characteristic and
channel models for
massive MIMO
dynamic channels
under various railway
scenarios at 6 GHz,
28 GHz, 38 GHz,
60 GHz and
300 GHz
frequency bands.
AI_LAYOUT.qxp_Author Layout 10/1/15 3:38 PM Page 81
IEEE Communications Magazine • October 2015
82
to intensive users simultaneously inside the car
with flexible grouping of hundreds of antennas.
Figure 3 shows the diagram of wireless coverage
for railways based on massive MIMO techniques.
Figure 3 denotes the signal access mode into
the car. The blue line denotes the signal at fre-
quency bands below 6 GHz used for wide area
coverage. However, 7.2 GHz bandwidth may be
demanded for HD video services. MmWave or
sub-mmWave bands such as 28 or 300 GHz
offers orders of magnitude greater spectrum
than current wireless allocations. They enable
high-dimensional antenna arrays for further
gains via beamforming and spatial multiplexing.
Moreover, the antenna size can be dramatically
reduced compared to that of the low frequency
bands. The red lines denote mmWave or sub-
mmWave.
There are mainly three access modes for the
coverage inside the cars. Just as depicted in Fig.
3 (1-1), the signal from the base station pene-
trates directly into the car with penetration loss
up to 24 dB, posing much higher requirements
for the transmission power of the base station
and the receiver sensitivity. When we use other
access modes, straight penetration of the signal
is unavoidable, which may cause interference to
useful signals. It is a challenge to figure out how
to shield the signal directly into the car when using
other access modes. Vehicle repeaters can be
adopted for high mobility environments, shown
as (1-2), the signals received from the on-vehicle
transceiver are modulated and forwarded to the
micro base or WiFi signal repeater. However,
the typical unresolved problem is that with a
repeater, an AP inside the car may receive a
weak signal through the train body and a strong
one from the repeater, but with considerable
delay. How to design transmission schemes for
good reception of a repeater at high moving speeds
is a challenging task. Figure 3 (1-3) represents a
two-hop access mode. One hop is from the base
station to the antennas on the top of the train,
and another is from antennas on top of the train
to receiver antennas inside the car. This may
avoid large penetration losses introduced by the
train body. However, as we know, higher fre-
quency bands have large attenuation and path
Figure 3. Massive-MIMO-based wireless coverage for railway systems.
Awning
Pylons
Complex
environment
Semi-closed
area
2
3
3.7 m
Millimeter
wave
25 m
Base station
massive MIMO
Base station
massive MIMO
20-30 m 20-30 m
1
Linear
Spherical
Rectangular
Distributed
Cylindrical
1-2
4-1
4-2
4
1-1
1-3
5
AI_LAYOUT.qxp_Author Layout 10/1/15 3:38 PM Page 82
IEEE Communications Magazine • October 2015 83
loss, resulting in very limited transmission range.
Therefore, the communication between the base
stations to the train (i.e., train-to-ground com-
munication) may use the working frequency
bands below 6 GHz. By using the multihop tech-
nique at different frequency bands [8], higher
frequency bands can be used inside the car for
large bandwidth wireless coverage.
Massive MIMO used in a dense urban area is
denoted by in Fig. 3. There may be many
users under this scenario, where massive MIMO
can be utilized to enhance the system capacity.
Since it is inconvenient to set up base stations
with large-scale antennas, we can deploy an
alternative form of massive MIMO antennas at
the base station side utilizing nearby high build-
ings. The shape of the antennas can be diverse
such as linear array, cylindrical array, spherical
array, plane array, and distributed antenna
arrays. The task here is that the channel models
and receiver design should be appropriate for vari-
ous shapes of antennas arrays.
Figure 3 represents the scenario in the rail-
way station with the characteristics of the semi-
closed scene, dense crowd, and complicated
environment. The deployment of the antennas is
restricted by the scenes; for example, the linear
array will probably not be equipped under such
a scenario. Therefore, how to design the appropri-
ate antenna array types, including the irregular
antenna array is an interesting task.
Figure 3 represents the scenario inside the
car. It needs to provide high data rate services
for many users simultaneously. The car is a
closed scene of rectangular shape, likely to cause
multiple backscattering, resulting in large attenu-
ations and losses, especially at mmWave or sub-
mmWave frequency bands. A 3D beamforming
technique can be utilized to overcome these
shortcomings. As for the detailed deployment of
a massive MIMO antenna array, the fundamen-
tal infrastructure such as the goods and luggage
racks inside the cars and the user distributions
should be considered as well, because they may
have effects on the shadow fading. The conven-
tional shadowing loss caused by the users for
indoor mmWve coverage can come to 4–5.5 dB.
Because the passengers are generally uniformly
distributed, it is the best choice to deploy anten-
nas with low power in a distributed form,
equipped in the center of the top of the car, ver-
tical to the train running direction. This may
either be applicable to the passenger distribution
or avoid the obstruction to the signals to a great
limit. In this way, the diversity gain can be fully
exploited. How to deploy the antennas array
including the antenna numbers, the shapes, and
the pitch angles efficiently to satisfy various scenar-
ios and requirements is a research direction for
massive MIMO.
In fact, the antenna can be either centrally
configured in a base station to form a central-
ized antenna array, as depicted by (4-1) in Fig. 3,
or configured among multiple nodes to form the
distributed antenna array, as depicted by (4-2) in
Fig. 3. The centralized antenna array provides
larger array gain and more spatial degree of
freedom to suppress intra- and inter-cell inter-
ference. The centralized antenna array may be
equipped on the top of the train, forming strong
propagation beams, to track the departure and
arrival of the train with the centralized power.
The installation of the centralized antenna can
use the directional mode, similar to the splayed
shape, to reduce the effect of Doppler shift. The
research on centralized massive MIMO applica-
tions at high mobility is a most challenging task.
The correlation coefficient of massive MIMO is
large, especially for centralized antennas. The
correlation and coupling effects may degrade its
performance dramatically. One research direc-
tion is analysis of the correlation and coupling
mechanisms for massive MIMO.
For distributed massive MIMO, there are
many antennas sufficiently close within a very
limited area, and strong beamforming can be
achieved with the coordination of different
antennas. Because of the cuboid shape of the
train car and very dense crowds, the advantages
of the distributed antenna can be fully released
compared to the centralized antenna. Reason-
able deployment of the distributed antennas
inside the car should be determined according to
specific conditions of the cars and the passenger
distribution. Note that distributed massive
MIMO is different from a distributed antenna
system (DAS). There is only one antenna at the
remote access unit (RAU) for DAS, but tens to
hundreds of antennas for distributed massive
MIMO. In the latter case, one user terminal can
see all the distributed antennas simultaneously
with beamforming, while there is no beamform-
ing effect for DAS.
Inter-car communication is denoted by in
Fig. 3. Optical fibers and wireless communica-
tions can both be adopted for inter-car commu-
nications. However, optical fibers are not
recommended to connect communication nodes
because it may be expensive to wire a train for
network access, and rewiring may be needed
every time the train is reconfigured. Currently,
the main possible wireless connection forms for
inter-cars are WiFi, WiMAX, and dedicated
short-range communications (DSRC).
CHANNEL MODELING AND
SYSTEM-LEVEL MODELING
Whether for GSM-R, LTE-R, 5G-R, or
mmWave applications, the major prerequisite
condition is a thorough knowledge of the propa-
gation characteristics of the wireless channel.
Wireless channel modeling is the important basis
and essential means for communication network
planning and optimization, transmitter and
receiver design, and physical and upper layer key
techniques selection. Recently proposed new 3-
D channel models such as the TR36.873 model
[11] are a step into the right direction. The new
model allows the physically correct modeling of
channels by clustering scatterers, supporting 3D;
to realistically model line-of-sight (LOS) and
non-LOS (NLOS) depending on distance; and to
include antenna arrays even of large dimensions.
Nevertheless, particular aspects of train connec-
tions are not yet supported or still need to be
sufficiently parameterized. As for massive
MIMO channel measurement, most of the litera-
ture now focuses on the virtual antenna array
Optical fibers are not
recommended to
connect communica-
tion nodes because it
may be expensive to
wire a train for net-
work access, and
rewiring may be
needed every time
the train is reconfig-
ured. Currently, the
main possible wire-
less connection
forms for inter-cars
are WiFi, WiMAX,
and dedicated short-
range communica-
tions.
AI_LAYOUT.qxp_Author Layout 10/1/15 3:38 PM Page 83
IEEE Communications Magazine • October 2015
84
measurement in static condition, the fast channel
sounding techniques and dynamic channel param-
eters extraction for massive MIMO are indeed
very challenging works.
Moreover, a very difficult problem remains
for system-level modeling, in which abstraction
models are being derived in mathematical form
to offer simulation results in acceptable time or
even allow for explicit analytical solutions. For
such methods to work, it is important to validate
the abstraction steps by smaller transmission
units that are compared to more detailed link
level descriptions. Once the models agree, the
system-level models can easily be scaled to hun-
dreds of users, higher bandwidth and longer
transmit durations with high likelihood to cor-
rectly resemble such behavior [10]. The unre-
solved issues are handovers as high-speed trains
contain many users (an inter-city-express, ICE,
carries typically more than 500) that need to be
handed over to the next station in a very short
time. Currently, no accurate channel models are
set up for fast handovers of so many users. Clas-
sical hybrid automatic repeat request (HARQ)
methods [11] may not work as expected due to
high speeds. Theoretically, sufficient (quantized)
feedback information is required in order to
guarantee high data rate transmissions [12].
However, at high speeds such channel state
information is quickly outdated, and once the
speed exceeds 100 km/h there is no gain in feed-
back information. While users are being moved
from one antenna port to the next, they typically
observe a U-shaped attenuation profile in
between, in which the received power drops to a
minimum once the center between two antenna
ports is crossed. Smart scheduling can compen-
sate for such effects offering a high data rate on
average per user [13].
Above all, we should develop the appropriate
communication network architectures, channel
and system-level models, and different layer key
technologies such as massive MIMO to meet the
needs of high data rate, high spectrum, and high
energy efficiency of mobile communication sys-
tem for railways.
CONCLUSIONS
HSR is developing very quickly in many parts of
the world. Safety, reliability, high efficiency,
environmental friendliness, comfort and human-
ization are the goals of future HSR develop-
ments. Against this application background, we
propose network architectures and some chal-
lenging research directions including propaga-
tion characteristics, channel models, antenna
designs, and seamless network modes switching.
With the implementation of massive MIMO
techniques and higher working frequency bands
into the railway systems, even higher transmis-
sion rates and more reliable wireless transmis-
sions can be realized at extremely high speeds,
aided by careful designs that preserve the high
quality of user experience.
ACKNOWLEDGMENTS
This work was supported in part by the NSFC
under Grant 61222105, National 863 Project
under Grant 2014AA01A706, the State Key Lab
project under Grant RCS2014ZT11 and
RCS2014ZZ03, the Key Project of Chinese Min-
istry of Education under Grant 313006, the
NSFC under Grant U1334202, and the Natural
Science Base Research Plan in Shaanxi Province
of China under Grant 2015JM6320.
REFERENCES
[1] B. Ai et al., “Challenges Toward Wireless Communications
for High-Speed Railway,” IEEE Trans. Intelligent Transporta-
tion Systems, vol. 15, no. 5, Oct. 2014, pp. 2143–58.
[2] K. Guan, Z. Zhong, and B. Ai,“Assessment of LTE-R
Using High Speed Railway Channel Model,” 3rd Int’l.
Conf. Commun. and Mobile Computing, Qingdao,
China, Apr. 2011, pp. 461–64.
[3] Z. Zhong et al., Fundamental Theory of GSM-R Wireless
Networks, Beijing Jiaotong Univ. Publishing House,
June 2009.
[4] METIS project, Deliv. D1.1, “Scenarios, Requirements
and KPIs for 5G Mobile and Wireless System,” tech.
rep., Apr. 2013.
[5] HORIZON 2020 Work Programme 2014–2015 11.
Smart, Green and Integrated Transport Revised, EC
Decision C(2014)4995, July 2014.
[6] B. Ai et al., “Social Network Services for Rail Traffic
Applications,” IEEE Intelligent Systems, vol. 29, no. 6,
Dec. 2014, pp. 63–69.
[7] T. L. Marzetta, “Noncooperative Cellular Wireless with
Unlimited Numbers of Base Station Antennas,” IEEE
Trans. Wireless Commun., vol. 9, no. 11, Nov. 2010,
pp. 3590–3600.
[8] T. Kikkawa and Y. Sanada, “Wireless Relay Networks
Using Multiple Frequency Bands,” 2010 Int’l. Symp.
Info. Theory and Its Applications, Taichung, Taiwan,
Oct. 2010, pp. 203–08.
[9] 3GPP, Release 12, V12.0.0, “Technical Specification
Group Radio Access Network: Study on 3D Channel
Model for LTE,” tech. rep., Sept. 2014.
[10] S. Schwarz, et al., “Pushing the Limits of LTE: A Survey
on Research Enhancing the Standard,” IEEE Access, vol.
1, May 2013, pp. 51–62.
[11] J. Ikuno, C. Mehlfuhrer, and M. Rupp, “A Novel Link
Error Prediction Model for OFDM Systems with HARQ,”
IEEE ICC, Kyoto, Japan, June 2011, pp. 1–5.
[12] S. Schwarz, and M. Rupp, “Predictive Quantization on
the Stiefel Manifold,” IEEE Signal Processing Lett., vol.
22, no. 2, Feb. 2015, pp. 234–38.
[13] S. Schwarz, C. Mehlfuhrer, and M. Rupp, “Throughput
Maximizing Multiuser Scheduling with Adjustable Fair-
ness,” IEEE ICC, Kyoto, Japan, June 2011, pp. 1–5.
BIOGRAPHIES
BOAI[SM] (aibo@ieee.org) received his Ph.D. from Xidian
University. He is a professor and Ph.D. advisor at Beijing
Jiaotong University. He is deputy director of the State Key
Lab of Rail Traffic Control and Safety in China. He has
authored and co-authored six books and more than 220
papers. He is an IET Fellow. His research interests are
focused on rail traffic mobile communications and channel
modeling.
KEGUAN [M] received his Ph.D. from Beijing Jiaotong Uni-
versity. He is an associate professor at Beijing Jiaotong Uni-
versity. He received the International Union of Radio
Science Young Scientist Award in 2014. He is a member of
the IC1004 initiative. He has authored and co-authored
more than 60 papers. His research interests include the
measurement and modeling of wireless propagation chan-
nels, rail traffic communications, and future terahertz com-
munication systems.
MARKUS RUPP [F] (mrupp@nt.tuwien.ac.at) received his Dr.-
Ing. degree at the Technische Universität Darmstadt, Ger-
many. Since 2001 he is a full professor at the Vienna
University of Technology. He served as dean from
2005–2007 and as head of the Institute from 2014–2015.
He has authored and co-authored more than 450 scientific
papers and patents on adaptive filtering, wireless commu-
nications, and rapid prototyping, as well as automatic
design methods.
THOMAS KÜRNER [SM] (kuerner@ifn.ing.tu-bs.de) received his
Dr.-Ing. Degree from Universität Karlsruhe, Germany. Since
With the implemen-
tation of massive
MIMO techniques
and higher working
frequency bands into
the railway systems,
even higher trans-
mission rates and
more reliable wire-
less transmissions
can be realized at
extremely high
speeds, aided by
careful designs that
preserve the high
quality of user
experience.
AI_LAYOUT.qxp_Author Layout 10/1/15 3:38 PM Page 84
IEEE Communications Magazine • October 2015 85
2003, he has been a professor of mobile radio systems at
TU Braunschweig, Germany. His work areas are propaga-
tion, self-organization of cellular networks, car-to-x com-
munications, and channel characterization for future
terahertz communication systems. He chairs the IEEE802.15
TG 100G and the WG of the European Association on
Antennas and Propagation.
XIANG CHENG [SM] (xiangcheng@pku.edu.cn) received his
Ph.D. degree from Heriot-Watt University and the Universi-
ty of Edinburgh, United Kingdom. He is an associate pro-
fessor at Peking University. He has authored and
co-authored over 80 papers. His research interests include
mobile propagation channel modeling and simulation,
next-generation mobile cellular systems, intelligent trans-
port systems, and hardware prototype development and
practical experiments.
XUE-FENG YIN [M] (yinxuefeng@tongji.edu.cn) received his
Ph.D. degree in wireless communications from Aalborg
University, Denmark, in 2006. Since 2008 he has been an
associate professor at Tongji University. He has published
more than 60 technical papers and co-authored a book on
channel characterization and modeling. His research inter-
ests are in parameter estimation for radio channels, chan-
nel characterization, and stochastic modeling.
QIWANG [S] received his B.S. degree in communication
engineering from Beijing Jiaotong University. He is current-
ly working toward his Ph.D. degree in the State Key Lab of
Rail Traffic Control and Safety, Beijing Jiaotong University.
His research interests mainly include channel modeling for
massive MIMO and vehicular-to-vehicular communications.
GUO-YUMAreceived his B.S. degree in electrical engineer-
ing from Beijing Jiaotong University, China, in 2012. Now
he is working toward his Ph.D. degree at the State Key Lab
of Rail Traffic Control and Safety, Beijing Jiaotong Universi-
ty. His current research interests are focused on massive
MIMO and non-orthogonal multiple access techniques for
5G.
YAN LI[S] received her B.S. degree in electrical engineering
in 2011 from Beijing Jiaotong University, where she is cur-
rently working toward her Ph.D. degree with the State Key
Lab of Rail Traffic Control and Safety. She is now studying
at the University of British Columbia, Canada, as an inter-
national exchange Ph.D. student. Her research interests are
in the field of measurement and modeling of wireless
propagation channels.
LEI XIONG received his Ph.D. from Beijing Jiaotong University
in 2007. He is an associate professor at Beijing Jiaotong
University. He has authored and co-authored two books
and more than 30 papers. He is an expert on railway com-
munications in China. His research interests are focused on
rail mobile communications, channel simulation, and soft-
ware defined radio.
JIAN-WEN DING received his M.S. degree from Beijing
Jiaotong University. He is a lecturer at Beijing Jiaotong
University. He is deputy director the of Laboratory of
Rail Traffic Mobile Communication of Beijing Jiaotong
University. He has authored and co-authored four books
and more than 20 papers. His research interests are
focused on rail traffic mobile communications and chan-
nel modeling.
AI_LAYOUT.qxp_Author Layout 10/1/15 3:38 PM Page 85
