High-speed Railway 2 (2024) 248–258

Contents lists available at ScienceDirect

High-speed Railway
journal homepage: www.keaipublishing.com/en/journals/high-speed-railway

Research article

WiFi performance analysis in high-speed railway communication

Ziqi Zhanga, Fengye Hua, Zhuang Linga,b, , Cong Liua, Fengting Xua
⁎ ]]
]]]]]]
]]

a
College of Communication Engineering, Jilin University, Changchun 130012, China
b
Beijing Engineering Research Center of High-speed Railway Broadband Mobile Communications, Beijing Jiaotong University, Beijing 100044, China

A R T I C L E I N F O A B S T R A C T

Keywords: In High-Speed Railways (HSRs), the Train Control and Management System (TCMS) plays a crucial role.
High-speed railway However, as the demand for train networks grows, the limitations of traditional wired connections have become
Train wireless communication network apparent. This paper designs and implements a Wireless Train Communication Network (WTCN) to enhance the
WiFi existing train network infrastructure. To address the challenges that wireless communication technology faces in
Communication performance measurement
the unique environment of high-speed rail, this study first analyzes various onboard environments and simulates
several typical scenarios in the laboratory. Integrating the specific application scenarios and service char­
acteristics of the high-speed train control network, we conduct measurements and validations of WiFi perfor­
mance, exploring the specific impacts of different factors on throughput and delay.

1. Introduction technologies such as Multiple-In Multiple-Out (MIMO) and Orthogonal

Frequency Division Multiplexing (OFDM) have been applied, giving
1.1. Background WiFi communication the characteristics of high bandwidth and low
delay. Furthermore, wireless transmission can offer intelligent man­
Since 1946, when the world’s first high-speed railway began op­ agement and control, aligning with the demands of certain industrial
eration, over 70 years have passed. During these decades, the train scenarios and thus being applied across various industrial fields
industry has undergone continuous development and transformation. [13,14].
High-speed operation is no longer the sole benchmark [1]; comfort and Compared to wired transmission, wireless network control can help
convenience have begun to inject new life and vitality into the railway high-speed train operators perform timely maintenance and repairs
industry [2–4]. The increase in the number of controllable devices in­ more efficiently, reducing wear and maintenance costs. Furthermore, it
side trains has led to a rise in the amount of data transmitted within contributes to improving the intelligence of trains, enabling more
train control systems. Simultaneously, with the ongoing evolution of flexible management and control, and laying the technical foundation
the railway industry, the demand for higher control precision and faster for the implementation of next-generation high-speed railway tech­
response times has driven experts and scholars to seek new data nologies, such as virtual coupling. A significant European railway re­
transmission methods. From the early LonWorks bus [5] to the more search and development initiative, Shift2Rail, aimed at driving tech­
well-known Control Area Network (CAN) bus [6] and Multifunction nological innovation and competitiveness in rail transport. One of the
Vehicle Bus (MVB) [7,8], data transmission methods have continuously purposes of the Shift2Rail Technologies for Autonomous Rail Operation
evolved. However, all of the aforementioned transmission methods rely (TAURO) project is to define and propose standardized remote driving
on cables, requiring the use of a large amount of wiring inside trains. and command functionalities for this type of operation in the future
This makes routine maintenance of equipment cumbersome and makes [15]. Roll2Rail, a precursor project to Shift2Rail, focused on developing
it difficult to identify the exact point of failure when malfunctions key technologies and resolving identified obstacles to foster radical
occur. The addition of new equipment requires re-cabling, which, to innovation in railway vehicles, aiming to increase operational relia­
some extent, hinders the intelligent development of trains. In contrast, bility and reduce Life-Cycle Costs (LCCs). The project explored Train
wireless transmission has developed rapidly in recent years Control and Management Systems (TCMS) using wireless technology for
[9–12]. Since the introduction of WiFi 4 based on IEEE 802.11n, control and monitoring functions, simplifying train coupling procedures

⁎
Corresponding author.
E-mail address: [email protected] (Z. Ling).

https://doi.org/10.1016/j.hspr.2024.11.005
Received 14 October 2024; Received in revised form 10 November 2024; Accepted 15 November 2024
Available online 20 December 2024
2949-8678/© 2024 The Authors. Published by Elsevier B.V. on behalf of KeAi Communications Co. Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Z. Zhang, F. Hu, Z. Ling et al. High-speed Railway 2 (2024) 248–258

Fig. 1. The channel links within the scope of the Shift2Rail project.

and eliminating the need for onboard communication cables. Devices system is integrated without affecting the existing communication
using wireless Train Communication Networks (TCNs) can be placed in system of the CR400 EMU, allowing it to serve as auxiliary commu­
the same vehicle, the same consist, different consists, or even different nication in case of wired connection failures. Furthermore, network
trains [16] (see Fig. 1). performance monitoring equipment is introduced in the experiment to
Although the introduction of wireless networks can enhance the in­ evaluate network performance in real-time, ensuring that the wireless
telligence and service level of high-speed trains, it also brings certain system can seamlessly take over and provide reliable communication
drawbacks [17]. Wireless communication is often subject to interference, to assist the wired communication system.
which can lead to fluctuations in delay and impact control performance • When analyzing the characteristics of different train scenarios, we
[18]. For example, the early adoption of WiFi (IEEE802.11) in the train observe significant differences in structural layout, signal propagation
industry appeared in metro systems through the Communication-Based conditions, and physical environments in areas such as inside the
Train Control (CBTC) system. The widespread use of other devices based carriages, between carriages, on the roof, and beneath the train. To
on the IEEE 802.11 protocol has exacerbated the 2.4 GHz co-channel further explore the performance of the wireless communication
interference problem faced by CBTC systems [19]. Therefore, before system in these scenarios, we select several representative environ­
deploying wireless train communication on trains, further research into ments for detailed analysis and evaluation. Additionally, we recreate
the performance of specific technologies is required, along with an the key features of these scenarios in a laboratory setting, providing
analysis based on empirical data. critical experimental data and valuable references for the design of
wireless communication systems in actual train environments.
1.2. Contribution and paper structure • In a laboratory-simulated train environment, we set up a complete
measurement system to evaluate the performance of WiFi technology
In this paper, a Wireless Train Communication Network (WTCN) is under specific conditions. These scenarios include the real-world effects
designed without affecting the existing wired communication network of of multipath propagation, signal attenuation, and the shielding effect
the CR400 Electric Multiple Units (EMUs). The key feature of this net­ caused by carriage partitions on wireless signal transmission. In this
work is that, in addition to enabling communication between different simulated environment, we deploy two reliable WiFi bridges available
carriages of the EMU, it incorporates a sensing system that can detect the on the market, each measured separately, allowing us to record real-
internal and external conditions of the train using various sensors, such time performance data of the WiFi network across different scenarios.
as LoRa and WiFi. To explore the characteristics of the train’s wireless
communication environment, an analysis is conducted based on different The rest of this paper is organized as follows. In Section 2, we in­
communication distances and locations of communication devices. troduce the current wired communication network topology of the CR400
Moreover, after analyzing several common wireless technologies, we intelligent EMU and based on this design a wireless communication net­
systematically evaluate the performance of WiFi in terms of two key work system. In Section 3, through the analysis of the wireless commu­
metrics, throughput and delay, by simulating wireless communication nication environment of the train, we categorize the communication into
scenarios inside high-speed rail carriages in a laboratory environment. three types: within a single carriage, between different carriages in the
This study investigates WiFi’s suitability for WTCN. The two WiFi bridges same consist, and between carriages in different consists. We also discuss
selected for this study are both well-established wireless communication the communication characteristics on the roof, inside the carriages, and
devices that are commercially available: one is a consumer-grade bridge, beneath the train. In Section 4, we first conduct a brief analysis of several
and the other is an industrial-grade bridge. The industrial-grade bridge common wireless communication technologies and select WiFi as the
focuses more on ultra-high bandwidth, security, and interference re­ focus. Then, we simulate several typical train scenarios in a laboratory
sistance. These devices were deployed and measured in a physical la­ environment by replicating real-world conditions such as corridors, metal
boratory that simulated a real train environment to obtain accurate doors, different heights, and varying distances for bridge placement. Based
network performance data. We present actual measurement results of on these replicated conditions, we built a measurement system to gather
delay and throughput, considering TCMS data and various in-train sce­ performance data on WiFi technology in different scenarios.
narios such as the inner end door, rather than relying on simulation re­
sults. Our contributions are summarized as follows: 2. A WTCN based on the CR400 Train Network System

• By analyzing the current train system of the CR400, we identify sui­ The CR400 EMU is an important representative of China’s high-
table locations to install wireless communication devices and network speed railway and belongs to the “Fuxing” series of high-speed trains. It
performance monitoring equipment. The wireless communication consists of eight carriages, with every four carriages forming a traction

249
Z. Zhang, F. Hu, Z. Ling et al. High-speed Railway 2 (2024) 248–258

unit, which includes one traction carriage and three regular carriages. long-distance, low-power data exchange between different train sets,
Thus, the eight carriages are divided into two traction units. particularly for transmitting information without relying on train-to-
Additionally, two EMU sets can be coupled together to operate jointly. ground communication. The LoRa network monitors track and en­
The TCN is structured into three layers: the Ethernet Train Backbone vironmental conditions using wireless sensor nodes such as wireless
(ETB), the Ethernet Consist Network (ECN), and the Ethernet Device infrared axle temperature sensors, wireless vibration sensors, wireless
Network (EDN). In the ETB, each carriage is equipped with an Ethernet temperature and humidity sensors, and wireless pressure and altitude
Train Backbone Node (ETBN) connected to the backbone. At the consist sensors, enhancing the safety and monitoring capabilities of the train’s
level, the ECN is used, with each carriage featuring two Ethernet operations [20]. The GNSS module provides accurate timing, posi­
Consist Network Nodes (ECNNs), one acting as the primary switch and tioning, and speed information, ensuring time synchronization and
the other as a backup, providing hot redundancy to ensure continuous positioning accuracy within the train network system, and providing
communication. These two switches, installed in the same location, are robust support for vehicle-level and train-level data transmission
connected to the carriage devices using a point-to-point star topology. [21,22]. WiFi APs cover the area beneath the train, on the roof, and
They share a virtual Media Access Control (MAC) address and Internet inside the carriages, offering stable access services for handheld term­
Protocol (IP) address, allowing the system to automatically switch to inals used by train staff and mobile devices used by passengers. Ad­
the backup switch in case of a failure in the primary switch, ensuring ditionally, a firewall is placed between the ETBN and ECNN to imple­
uninterrupted communication. To provide redundant links and ensure ment data isolation and access control. Multi-layer security measures,
the communication system can continue operating in case of a wired such as encrypted communication, authentication and authorization,
network failure, as well as to enhance flexibility between device-level intrusion detection and defense, network segmentation with Virtual
and consist-level networks and reduce reliance on physical wiring, we Local Area Networks (VLANs), and security protocols and standards, are
designed a WTCN based on the CR400 Train Network System. The to­ applied to enhance the system’s security and reliability. To compre­
pology is illustrated in Fig. 2. hensively monitor the performance of the train’s network, we have also
In the figure, the blue devices represent the existing equipment in introduced a network measurement system that specifically monitors
the CR400 TCMS, while the red devices are the newly designed com­ and analyzes the current network’s throughput and delay. For
ponents of the WTCN based on the CR400 Train Network System. The throughput monitoring, we use a Host Computer to monitor the data
ETB uses bridge devices to connect the ETBNs between different train flow in real-time. Through the Host Computer, we accurately track the
units, enabling wireless communication between the ETBNs of different data transmission rate within the network and record throughput per­
traction units. Additionally, within a single train, bridges are used to formance under different workloads. This helps us identify network
connect each train’s ECNN to the onboard equipment, establishing a bottlenecks during high-load conditions and provides detailed traffic
wireless communication system between the onboard devices and the statistics for system optimization and troubleshooting. For delay mon­
ECNN of each carriage. We have also introduced a LoRa communication itoring, to ensure accuracy, we design and deploy a dedicated delay
gateway, a Global Navigation Satellite System (GNSS) module, and measurement tool called the Net Delay Tester (NDT). This tool is cre­
WiFi Access Points (APs), while adding security measures between ated to avoid the potential delay errors caused by operating system lag,
the ETBN and ECNN. The LoRa communication gateway enables resource contention, or other external factors when using a Host

Fig. 2. WTCN for CR400.

250
Z. Zhang, F. Hu, Z. Ling et al. High-speed Railway 2 (2024) 248–258

Computer. With this system, we monitor the network’s response time in load management become significant challenges. The simultaneous con­
real-time, ensuring the low-delay characteristics of the train control nections of many users may lead to network congestion and reduced speeds.
system and quickly identifying and addressing potential delay issues. 3) Multipath effect in confined spaces. The enclosed, environ­
This ensures the overall stability and efficiency of the train’s commu­ ment of the train carriage amplifies the inevitable multipath effect. The
nication system. significance of this effect is related to the signal wavelength and the size
of obstacles, as signals may reflect, scatter, and refract when interacting
3. Scenario analysis in high-speed rail with objects, increasing the complexity of signal reconstruction.
4) Electromagnetic shielding loss. Communication within the
By studying wireless communication scenarios in high-speed trains, we metal-encased carriage can experience significant electromagnetic ab­
gain a deeper understanding of the challenges, enabling more effective sorption due to the conductivity of metal, which absorbs electromagnetic
solutions and exploration of various wireless communication technologies. waves. The free electrons in the metal, driven by the electric field, vibrate
We classify wireless communication in trains into three categories: com­ and absorb the energy of the electromagnetic waves. The absorption rate
munication within a single carriage, communication between carriages in is higher in metals with high electrical conductivity because they have
the same consist, and communication between different consists. These are more free electrons to interact with the waves. Factors such as the in­
further divided into three key areas for discussion: the roof, the interior of cident angle, thickness of the metal, and frequency of the electromagnetic
the carriage, and beneath the train. A detailed analysis of these scenarios waves also influence this absorption, requiring careful consideration to
provides valuable insights for optimizing the use of wireless communica­ ensure adequate transmission performance in specific applications.
tion technologies in high-speed trains, ensuring that reliability and per­
formance requirements are met in each specific environment. 3.1.3. Under-carriage equipment
If the communication device is placed under the train, the channel
3.1. Single carriage interferences under the train are as follows:
1) Strict delay requirements. Similar to roof-mounted equipment,
In the wireless communication environment within a single car­ the under-carriage environment demands low delay for control signal
riage, three key areas are considered: the roof, the interior, and the transmission. The frequent vibrations and multipath effects in this area
underside of the carriage. Each area involves terminal devices con­ present additional challenges to the real-time performance of wireless
necting to corresponding AP nodes, with data communication occurring communication.
through a bridge-based structure that links the AP nodes. Due to the 2) Vehicle vibration. The train experiences vibrations caused by
differences in channel characteristics and the types of data being track irregularities, rail joints, sleepers, and other factors.
transmitted in each area, it is essential to analyze the communication Accelerations, decelerations, turns, and braking further intensify the
technologies specifically for these carriage-level scenarios, including vibrations, such as lateral vibrations during turning. These vibrations,
the roof, the interior, and the underside of the train. though inevitable, affect the signal quality of the wireless commu­
nication system in high-speed trains.
3.1.1. Roof-mounted equipment 3) Multipath effect in the under-carriage environment. When
The interferences of the channel on the roof of the train are as fol­ signals are transmitted beneath the train, they are subject to reflections,
lows when the antennas installed on the roof of the train. diffractions, and scatterings from both the bottom of the train and the
1) Electromagnetic interference. The operation of high-voltage ground. This results in multipath interference, where the receiving end
power grids generates electromagnetic radiation and interference, experiences delayed signals due to multiple paths, leading to signal
which affect nearby wireless communication systems. The frequency distortion and delay extension.
range of the electromagnetic radiation from the power grid may overlap 4) Shielding loss in narrow spaces. The signals transmitted near
with the frequency bands used by wireless communication systems, the metal under-carriage are similarly affected by electromagnetic ab­
leading to signal interference that degrades the quality and reliability of sorption as in the carriage. Moreover, the confined space beneath the
communication. Additionally, switching operations, load changes, and train increases the likelihood of signal refraction and reflection, further
faults in the high-voltage grid can create transient interference, which intensifying the absorption effect from the metal structures.
includes broadband signals and sudden energy fluctuations, potentially
disrupting the normal operation of wireless communication systems. 3.2. Between carriages
2) Strict delay requirements. Due to the high-speed operation of
trains, the communication system must be capable of transmitting data The communication between carriages faces the following chal­
and responding quickly to ensure real-time communication. This is lenges:
particularly crucial for real-time interaction and data transmission re­ 1) Obstacle interference. Unlike the obstacles inside the carriage,
lated to train control signals. when the wireless transceiver bridges are located between carriages,
3) External environmental interference. High-speed trains pass the metal exterior of the train body further increases signal penetration
through various regions and environments during operation, which loss and shielding interference.
may expose the roof-mounted wireless communication systems to in­ 2) Greater relative movement of transceiver devices. The car­
terference from external sources such as radio stations, mobile com­ riages are connected via passageway mechanisms, and the relative
munication base stations, and radar equipment. These sources can in­ positions of the transceivers on either side of the connection change
terfere with wireless signal transmission on the roof of the train. continuously as the train operates. This relative movement interferes
with signal alignment and transmission.
3.1.2. In-carriage equipment 3) Dual requirements for throughput and delay. Communication be­
If the communication device is placed inside the train carriage, the tween carriages must account for both control information and data trans­
channel interferences inside the carriage are as follows: mission. Control signals between train carriages require low delay, while data
1) Obstructions. In the carriage, obstacles such as seats, luggage transmission requires high throughput. Therefore, communication between
racks, and passengers can disrupt the propagation of communication carriages must balance both data volume and delay requirements.
signals in this complex environment. The metal structures inside the 4) Strong electromagnetic environment. Similar to the electro­
carriage can further interfere with and attenuate signal transmission. magnetic interference experienced by roof-mounted equipment, the strong
2) High user density and network load. With a large number of electromagnetic environment continues to pose challenges to the normal
passengers accessing the network simultaneously, network capacity and operation of the wireless communication system between carriages.

251
Z. Zhang, F. Hu, Z. Ling et al. High-speed Railway 2 (2024) 248–258

5) Mechanical structure (antenna size constraints). When an­ optimized according to their specific environments and performance
tennas are mounted on the roof for inter-carriage communication, their requirements to achieve optimal results.
size must be carefully considered to avoid being too close to the over­
head power lines, which could lead to severe electromagnetic inter­ 4. WiFi measurement in various laboratory scenarios
ference. Additionally, when the train is operating at high speeds, the
antenna size needs to be optimized to minimize wind resistance, which We first conducted a simple evaluation and summary of the ap­
could disrupt communication. Moreover, the antennas must be durable plicability of several mainstream wireless communication technologies
and resistant to vibration, impact, temperature changes, and harsh for the WTCN. We analyzed their performance in terms of key metrics
weather conditions, imposing higher design requirements. The struc­ such as bit rate, delay, high throughput, and network scale, as well as
ture of the antenna will also influence signal propagation. their suitability for use in WTCN.
6) Flexible network structure. The communication equipment 1) ZigBee and WirelessHART. ZigBee does not meet the bit rate
between carriages can form a flexible network. If a node in the com­ requirements of WTCN. Although its delay is suitable for process data
munication network fails or is damaged, the network can automatically and monitoring data, the overall delay remains too high. WirelessHART
reconfigure itself to restore communication. This is achieved by lever­ exhibits even greater delay, making it significantly unsuitable for the
aging the redundancy of the antenna coverage to bypass the damaged delay demands of WTCN. Therefore, neither ZigBee nor WirelessHART
node and connect to more distant devices, ensuring reliable data can be considered suitable wireless technologies for WTCN.
transmission and connectivity. Therefore, when implementing com­ 2) Ultra-WideBand (UWB). UWB fails to meet the high throughput
munication technologies, it is important to consider the communication required for process data and video streams in WTCN. Additionally, UWB
performance under varying distances and network requirements. lacks a deterministic medium access control mechanism and is designed
only for short-distance coverage, making it unsuitable for WTCN.
3) Millimeter-wave. Millimeter-wave technology can provide ex­
3.3. Between coupled consists tremely high throughput but remains immature and is highly sensitive
to environmental obstacles. Thus, its practical application may face
The communication between coupled consists faces the following challenges, making it not fully suitable for WTCN’s needs, although
challenges: further measurement may be required.
1) Dual requirements for throughput and delay. Coupled train 4) WiFi (IEEE 802.11). WiFi is a widely used wireless communication
consists also face electromagnetic radiation generated by electrical technology across various network environments. Its main advantages in­
equipment and wiring, interference from external wireless signals, and clude high bit rates and large network coverage, making it capable of
multipath effects caused by the semi-enclosed spaces formed between meeting the needs of networks of varying scales. WiFi 6 (IEEE 802.11ax),
the coupled consists. Communication between coupled train consists the latest version of WiFi, introduces key improvements such as OFDM and
involves transmitting information across entire train formations, which Multi-User Multiple Input Multiple Output (MU-MIMO), which significantly
demands higher data volumes and stricter guarantees of communica­ enhance network capacity, efficiency, and device collaboration. The com­
tion quality. Therefore, it is essential to consider employing one or more munication system based on OFDM has become an efficient solution for
technologies that can simultaneously meet both performance metrics. achieving high data rates in mobile environments. In such environments,
2) Inability to share a common ground and strong common- multipath propagation of signals often leads to delays, which may cause
mode electromagnetic interference between trains. When high- Inter-Symbol Interference (ISI). To mitigate this ISI, the guard interval be­
speed train consists are coupled, differences in their power systems, tween OFDM symbols ideally ensures that the overlapping parts of the signal
electrical equipment, and physical structures prevent them from are minimized, thus maintaining the integrity of the transmitted data and
sharing a common ground. Common grounding in the power grid is improving the overall reliability of the communication system. In high-speed
crucial for ensuring consistent electrical potential between devices, rail environments, mobility and multipath propagation are major challenges,
which is key to establishing reliable signal transmission and reducing making this setup particularly advantageous in environments like high-
interference. Without common grounding, multiple ground loop paths speed railway communication. Based on these factors, we believe that WiFi
can form, leading to ground loop interference. Additionally, issues may be a viable solution for WTCN, but its suitability needs to be evaluated
with power grid grounding may increase electromagnetic interference through actual measurement. Therefore, we simulated the train environ­
between trains, affecting the normal transmission of wireless signals ment in a laboratory setting to conduct further measurement.
onboard. We focus exclusively on the end-to-end data transmission perfor­
3) Different mechanical structures (antenna size requirements). mance of WiFi technology, without considering internal data processing
The communication design between coupled train consists must also in train devices. As a result, a measurement system was set up in the
consider the impact of mechanical structural differences. Due to var­ laboratory environment, as shown in Fig. 3.
iations in environmental conditions, the communication designs be­ This system is composed of two host computer testing terminals, a
tween coupled consists and between individual carriages must be wireless bridge, and an NDT. The wireless bridge is treated as a black

Fig. 3. WiFi performance measurement system in laboratory environment.

252
Z. Zhang, F. Hu, Z. Ling et al. High-speed Railway 2 (2024) 248–258

Table 1
Device specifications.

Specification TL-S520 TL-XCPE4800G

Wireless standard IEEE 802.11a/n/ac IEEE 802.11a/n/ac/ax

Operating band 5 GHz 5 GHz
Theoretical wireless rate 867 Mbps 4804 Mbps
Maximum transmission power 21.5 dBm 26 dBm
Antenna gain Built-in 11 dBi high gain dual-polarized antenna Built-in 15 dBi directional antenna
Wireless coverage angle / Horizontal: 35∘, Vertical: 15∘
Product size 224 mm × 79 mm × 60 mm 241 mm × 230 mm × 63 mm
Dust and water resistance level IP55 Highest level IP68

box, receiving input based on the measurement cases and providing the potential bottlenecks and provide valuable data for further technical
measurement results. For the wireless bridge devices, we selected the optimization.
TL-S520 and TL-XCPE4800G models, both set to a transmission power
of 20 dB. The specific specifications for the two devices are listed in
Table 1. 4.1. Laboratory-simulated scenarios
In the network measurement system, delay measurement is con­
ducted using our designed NDT. Data is transmitted through the train In this study, we simulate typical train scenarios in a laboratory
network, and timestamps are applied at the application layer before environment, specifically focusing on wireless communication within
being sent to the Host Computer for calculation, allowing us to measure and between train carriages. To replicate the train’s operational en­
the network’s delay and jitter. NDT communicates using the Modbus vironment as realistically as possible, we choose some selected loca­
protocol, with the timestamps for both receiving and sending data lo­ tions such as corridors and open spaces to perform the simulations,
cated at the application layer of the Open Systems Interconnection emphasizing the propagation characteristics and interference factors of
(OSI) model. The time precision is at the microsecond level. The Master wireless signals in these enclosed or semi-enclosed spaces. In the si­
NDT actively initiates a data request to the Slave NDT. First, it allocates mulations of both intra-carriage and inter-carriage scenarios, we mea­
buffer space for the data, and then fills the buffer with data. After the sured the performance of wireless signals in different environments to
data request is sent, the buffer space is released. When sending the data evaluate the effectiveness and stability of signal coverage. Through
request, the Master NDT records the timestamp T1. Upon receiving the these experiments, we get a comprehensive understanding of the real-
data request from the Master NDT, the Slave NDT responds and sends a world performance of wireless communication technologies in train
reply. When the Master NDT receives the returned data, it records the environments, providing empirical data and technical support for fur­
timestamp T4. After the initial connection is established, the Slave NDT ther optimization. The specific simulation details are as follows: We
obtains the port number of the Master NDT. The Slave NDT records select a corridor to simulate the train environment and evaluate the
timestamp T2 when receiving the data request and records T3 when real-world performance of wireless communication technologies in si­
sending the reply. After processing the exported data and removing the milar settings. Corridors, as enclosed or semi-enclosed narrow spaces,
internal processing delay of the NDT, the formula for calculating the exhibit physical characteristics similar to those of train carriages, such
one-way transmission delay is as wall reflections, multipath effects, and signal attenuation. Initial si­
mulation results for the path loss model inside a train car are in good
= (T4 T1) (T3 T2) (1) agreement with ITU-R results for a corridor model (see Fig. 4) [23].
The following method is used for throughput measurement: Two
Personal Computers (PCs) are used as Host Computers, running 4.2. Laboratory-simulated corridor environment
IxChariot for traffic measurement. A connection is established between
the two PCs based on their IP addresses. The data volume for measuring First, a corridor environment can effectively simulate the wireless
is determined by the parameters set in the data scripts of the mea­ signal propagation conditions inside a train carriage. Since train car­
surement cases. The formula for calculating the total data volume is as riages are typically long and narrow enclosed spaces, signals are easily
follows: affected by multipath effects during transmission. The physical struc­
ture of a corridor can accurately reproduce this phenomenon, making
D = file_size × transactions × number_of_ records (2)
the measure results highly valuable as a reference. Second, the enclosed
In this equation, file_size represents the size of each data packet, nature of the corridor makes signal propagation more susceptible to
transactions refer to the number of times data is sent, indicating how interference and reflection, similar to the signal interference and re­
many file_size packets are included in each record, and number_o­ flections caused by the metal structure of the train and the high
f_records denotes the total number of records generated. To match the
characteristics of data transmission in train networks, the parameters
for this experiment are set as transactions = 1, file_size = 10 MB, and
number_of_records = 100.
In train networks, large volumes of data, such as video surveillance
data between carriages and status updates from onboard systems, often
need to be transmitted frequently. The size of these data packets typi­
cally ranges from several hundred kilobytes to several megabytes.
Therefore, setting FS = 10 MB aligns with the data transmission scale
commonly seen in practical applications. By sending data multiple
times, we simulate scenarios where large data packets are transmitted
over multiple concurrent connections, measuring the network’s capa­
Fig. 4. Train corridor simulation model.
city and stability. These rigorous measuring conditions help identify

253
Z. Zhang, F. Hu, Z. Ling et al. High-speed Railway 2 (2024) 248–258

Fig. 5. Corridor measure actual scenario.

passenger density during operation. The measure scenario is shown in

Fig. 5.
The measurement results in the laboratory-simulated environment
are as Fig. 6.
In the scatter plot of the delay measurement results, each point re­
presents an individual measurement. The red line and the adjacent
value indicate the mean of the dataset, while the gray area represents
the range from the mean minus the standard deviation to the mean plus
the standard deviation. The TL-S520, the measurement results show
relatively stable throughput throughout the entire measure period,
averaging around 390 Mbps. In terms of delay, the average delay is
350.93 μs, displaying a relatively small fluctuation range. The TL-
XCPE4800G, its throughput exhibited a very stable trend, consistently
maintaining around 950 Mbps. However, its average delay was 2173.33
μs. This higher delay is due to the larger inter-frame spacing configured
in the TL-XCPE4800G, which helps reduce network congestion in
complex environments and improves device compatibility.
Additionally, the TL-XCPE4800G utilizes the WPA3 security protocol,
providing stronger encryption and authentication mechanisms.

4.3. Laboratory-simulated ground-level WiFi environment

To simulate the impact of different AP placements on WiFi signal

Fig. 6. Corridor measurement results.
performance inside the train and to explore WiFi signal behavior under
the train, we conduct measurements with the wireless bridge placed
near the ground. By comparing these results with the previous data, we 4.4. Laboratory-simulated inner end door environment
observe how different AP placements affect signal performance. This
analysis provides effective references for understanding the waveguide A metal door was used in the corridor simulation to represent the
effect in trains, helping to optimize signal propagation and improve inner end doors between train carriages in high-speed trains, aiming to
overall network performance in train carriage environments. As seen in evaluate the performance of wireless communication technologies be­
Fig. 7. tween carriages. The physical dimensions and structure of the metal
The measurement results in the laboratory-simulated ground-level door closely resemble those of the inner end doors in high-speed trains,
WiFi environment as seen in Fig. 8. effectively simulating the physical transition environment between
In our Ground-level measurement, we analyze the performance of carriages. In high-speed trains, the inner end doors are not merely
the TL-S520 and TL-XCPE4800G devices in simulated ground-level passageways; when they are closed, they can significantly impact signal
(close-to-ceiling) conditions. The TL-S520 shows an average throughput obstruction, reflection, and attenuation. By simulating this structure,
of around 390 Mbps during testing, though some fluctuations are ob­ we identify and analyze these issues in advance. The inner end door and
served. Its delay is primarily concentrated between 200 and 500 μs, the metal door used for the simulation are shown in Fig. 9.
with an average delay of 346.684 μs, demonstrating good response The measurement results in the laboratory-simulated inner end door
times. The TL-XCPE4800G exhibited outstanding throughput perfor­ environment as seen in Fig. 10.
mance, consistently maintaining around 950 Mbps, with an average The TL-S520 experienced a significant decrease in throughput and
delay of 2240.31 μs. This performance is attributed to the use of di­ exhibited noticeable fluctuations, with large differences between peak
rectional antennas in the TL-XCPE4800G, which helps reduce lateral and valley values, reflecting instability in signal transmission. The main
signal scattering. throughput fluctuated between 150 and 210 Mbps. Its delay was

254
Z. Zhang, F. Hu, Z. Ling et al. High-speed Railway 2 (2024) 248–258

Fig. 7. Ground-level measure actual photo.

primarily concentrated between 200 and 600 μs, with an average delay
of approximately 351.25 μs. The TL-XCPE4800G also showed a sig­
nificant decrease in throughput and evident fluctuations. The average
delay reached as high as 2139.57 μs. These data indicate that the
physical properties of the metal door have a significant impact on WiFi
signals.

4.5. Open-area simulation of wireless communication between traction

carriages

To simulate the wireless communication between Car 1 and Car 8

(the two traction carriages), we conducted measurements by placing
two wireless bridges 200 m apart in an open area. The model is shown
in Fig. 11.
This setup aims to simulate the communication distance between
the two furthest carriages within the train. By conducting this long-
distance communication measurement, we effectively verify the ap­
plicability of bridge devices between carriages, assess the impact of
distance on WiFi signal transmission, and provide valuable reference
data for actual deployment.
The measurement results from the open-area simulation of wireless
communication between the traction carriages as seen in Fig. 12. Fig. 8. Ground-level measurement results.
The TL-S520 experienced a significant decrease in throughput and
was highly unstable, with peaks reaching up to 28 Mbps and valleys
dropping to around 8 Mbps. This suggests very poor signal performance
over long-distance transmission. In terms of delay, most values were
concentrated between 500 and 2500 μs, with an average delay of ap­
proximately 926 μs, indicating substantial fluctuations during long-
distance measurements. The TL-XCPE4800G showed relatively stable
throughput at around 150 Mbps, demonstrating better data transmis­
sion capability compared to the TL-S520. However, its average delay
reached 3372.52 μs.
To further analyze the delay performance of the TL-S520 and TL-
XCPE4800G across different scenarios, we present the following two
comparison charts. Fig. 13 shows the delay comparison chart of TL-
S520 in various scenarios and Fig. 14 shows the delay comparison chart
of TL-XCPE4800G under the same conditions.
In the current TCN of the CR400, the maximum bandwidth of a
single network line is limited to 100 Mbps, with a transmission interval
between network segments of less than 20 ms. By analyzing the actual
performance test results of WiFi technology in a laboratory setting, we
found that the multipath effect inside the train carriage is very pro­
minent in HSR environments and has a significant impact on the signal. Fig. 9. Comparison of train door and laboratory door.

255
Z. Zhang, F. Hu, Z. Ling et al. High-speed Railway 2 (2024) 248–258

Fig. 12. Open-area measurement results.

Fig. 10. Door measurement results.

Fig. 11. Open-area simulation model.

However, WiFi 6 performs excellently in terms of throughput, and its

delay is also within the required limits. Compared to the TL-S520, the
TL-XCPE4800G not only offers higher throughput due to its bandwidth
but also provides greater stability, indicating that the TL-XCPE4800G
with WiFi 6 is better equipped to handle multipath propagation in HSR Fig. 13. The delay comparison chart of TL-S520 in different environments.

256
Z. Zhang, F. Hu, Z. Ling et al. High-speed Railway 2 (2024) 248–258

antennas also helps reduce the impact of multipath propagation.

Additionally, as there are numerous devices on the train that need to
connect to the ECNN, technologies such as MIMO and beacon slotting
will further assist in improving communication reliability.
Our future work will involve measuring additional wireless com­
munication technologies and refining the experimental setup, while
also examining other factors that could impact signal performance.
Additionally, we plan to conduct field measurements by integrating the
designed WTCN with onboard devices. Building on the preliminary
exploration of the waveguide effect within trains as presented in Ref.
[24], our research will further investigate its practical applications and
potential benefits. Furthermore, we will refine wireless communication
protocols to ensure they are specifically tailored to meet the unique
requirements of high-speed railways.

Fig. 14. The delay comparison chart of TL-XCPE4800G in different environ­ Declaration of Competing Interest
ments.
The authors declare the following financial interests/personal re­
lationships which may be considered as potential competing interests:
environments. However, it still faces challenges from the more complex Fengye Hu reports financial support was provided by the National
real-world environments on actual trains, particularly the impact of Natural Science Foundation of China. Zhuang Ling reports financial
metal objects on the signal. support was provided by Beijing Engineering Technology Research
Center of High-speed Railway Broadband Mobile Communication of
5. Conclusions Beijing Jiaotong University. If there are other authors, they declare that
they have no known competing financial interests or personal re­
In this paper, we studied the problem of WTCNs in High-Speed lationships that could have appeared to influence the work reported in
Railways (HSRs). Firstly, we designed a WTCN based on the CR400 this paper.
TCN. The WTCN not only enables wireless communication between
carriages but also incorporates a sensing system that can detect internal Acknowledgements
and external states using various sensors like LoRa and WiFi. To better
explore the application of different wireless communication technolo­ The authors would like to acknowledge support from the Beijing
gies, we investigated the characteristics of the communication en­ Engineering Research Center of High-speed Railway Broadband Mobile
vironment between different carriages in high-speed EMUs and ana­ Communications (BHRC-2024–1), Beijing Jiaotong University, and the
lyzed the challenges encountered. We ensured that the laboratory- National Natural Science Foundation of China (U21A20445).
simulated environment and physical dimensions closely resemble rea­
lity. By analyzing these issues and several common wireless commu­ References
nication technologies, we ultimately selected WiFi technology for
[1] F. Hu, Z. Ling, T. Liu, et al., Wireless perception of high-speed railway commu­
measurement. We simulated several typical scenarios in a laboratory
nication: Challenges, framework, and future directions, IEEE Wirel. Commun. 31
environment, built a measurement system, and using two devices, (2024) 284–292.
conducted measurements to explore the effects of different configura­ [2] N. Prasad, S. Jain, Intelligent Technologies in High-Speed Rail Transit Systems,
tions in the WiFi band for various HSR environments. Through actual Springer Nature Singapore, Singapore, 2024, pp. 217–249.
[3] G. Zhong, K. Xiong, Z. Zhong, et al., Internet of things for high-speed railways,
measurement data, we studied the propagation characteristics in the Intell. Converg. Netw. 2 (2021) 115–132.
WiFi band for HSR environments. Through these measurements, we [4] T. Liu, F. Hu, Z. Ling, C. Li, Y.C. Liang, Multi-agent cooperation-based deep re­
were able to better understand the possibilities and potential perfor­ inforcement learning for multisensor perception communication system in hsr
tunnel scenario, IEEE Trans. Commun. (2023) 1.
mance of using WiFi technology for wireless communication in high- [5] Z.J. Yu, H.M. Shi, L. Ai, et al., Lonworks-based intelligent train’s fire alarming
speed trains, providing empirical data and valuable references for fu­ control network, The 2nd International Workshop on Autonomous Decentralized
ture network design and optimization. Additionally, we found that System, IEEE, Beijing, 2002, pp. 262–266.
[6] S. Yong, Y. Ma, Y. Zhao, et al., Analysis of the Influence of Can Bus Structure on
while using the Wi-Fi Protected Access 3 (WPA3) security protocol Communication Performance, in: B. Li, J. Zheng, Y. Fang, et al. (Eds.), Springer
enhances data protection, it also increases processing time due to the International Publishing, Cham, 2020, pp. 405–416.
complex encryption and decryption processes, which subsequently [7] J. Feng, X. Lu, W. Yang, et al., Survey of development and application of train
communication network, in: L. Jia, Z. Liu, Y. Qin, et al. (Eds.), Proceedings of the
leads to increased delay. In situations involving high data traffic, the
2015 International Conference on Electrical and Information Technologies for Rail
more complex security handling can result in further delays. Transportation, Springer Berlin Heidelberg, Berlin, Heidelberg, 2016, pp. 843–854.
In our experiments, we observed that inner end doors significantly [8] J. Jimenez, J.L. Martin, U. Bidarte, et al., Design of a master device for the multi­
function vehicle bus, IEEE Trans. Veh. Technol. 56 (2007) 3695–3708.
impact the throughput of wireless networks, primarily because they
[9] D. Jia, F. Hu, Q. Zhang, et al., Distributed deep reinforcement learning-based power
physically obstruct the direct transmission paths of signals, thus directly control and device access for high-speed railway networks with symbiotic radios,
reducing transmission efficiency. Despite some influence on network IEEE Trans. Commun. (2024) 1.
delay from inner end doors, as long as the system is equipped with [10] Z. Li, F. Hu, Q. Li, et al., Aoi-aware waveform design for cooperative joint radar-
communications systems with online prediction of radar target property, IEEE
efficient error correction and data retransmission mechanisms, it can Trans. Commun. (2024) 1.
maintain low delay, effectively mitigating the impact of signal at­ [11] X. Yao, F. Hu, Z. Ling, et al., Robust beamforming design for ios-assisted multi-user
tenuation on data packet loss. miso systems with imperfect csi, IEEE Internet Things J. (2024) 1.
[12] D. Jia, F. Hu, Z. Ling, et al., Aoi-aware power control and subcarrier assignment in
In long-distance communication, directional antennas exhibit ex­ d2d-aided underlaying cellular networks for high-speed railways, IEEE Trans. Intell.
cellent performance, which is beneficial for re-establishing commu­ Transp. Syst. 24 (2023) 14237–14253.
nication between carriages when a problem occurs in one carriage, as [13] J. John, M. Noor-A-Rahim, A. Vijayan, et al., Industry 4.0 and beyond: The role of
5g, wifi 7, and time-sensitive networking (tsn) in enabling smart manufacturing,
well as for ensuring reliable communication between two train units Future Internet (2024) 16.
during coupling. In a carriage environment, the use of directional

257
Z. Zhang, F. Hu, Z. Ling et al. High-speed Railway 2 (2024) 248–258

[14] J. Du, W. Xiong, J. Wang, et al., A heuristic AP deployment approach for industrial [20] J. Soares, M. Luís, S. Sargento, et al. Mobile LoRa gateway for communication and
wireless networks, 2021 China Automation Congress, IEEE, Beijing, 2021, pp. 8035–8040. sensing on the railway, 2023 IEEE Symposium on Computers and Communications,
[15] J. Goikoetxea, I. de Arriba, I. Lopez, et al., Remote driving and command of trains: IEEE, Gammarth, 2023, pp. 499–502.
The shift2rail approach, Transp. Res. Procedia 72 (2023) 3723–3729. [21] J. Beugin, C. Legrand, J. Marais, et al., Safety appraisal of gnss-based localization systems
[16] J.M. Garcia-Loygorri, J. Goikoetxea, E. Echeverria, et al., The wireless train commu­ used in train spacing control, IEEE Access 6 (2018) 9898–9916.
nication network: Roll2rail vision, IEEE Veh. Technol. Mag. 13 (3) (2018) 135–143. [22] S. Spinsante, C. Stallo, Hybridized-gnss approaches to train positioning: Challenges
[17] Z. Ling, F. Hu, T. Liu, et al., Z. Han, Hierarchical deep reinforcement learning for and open issues on uncertainty, Sensors 20 (7) (2020) 1885.
self-powered monitoring and communication integrated system in high-speed [23] J. Elizalde, A. Arriola, M. Roset, et al. Design and validation of a wireless network
railway networks, IEEE Trans. Intell. Transp. Syst. 24 (2023) 6336–6349. for intra-train communications, 2024 18th European Conference on Antennas and
[18] J. Moreno, J.M. Riera, L.d. Haro, et al., A survey on future railway radio communications Propagation, IEEE, Glasgow, 2024, pp. 1–5.
services: challenges and opportunities, IEEE Commun. Mag. 53 (2015) 62–68. [24] N. Kita, T. Ito, S. Yokoyama, et al., Experimental study of propagation character­
[19] X. He, J. Li, Y. Wang, et al., Research on co-frequency interference of CBTC signal istics for wireless communications in high-speed train cars, 2009 3rd European
based on IEEE 802.11, 2018 IEEE 4th International Conference on Computer and Conference on Antennas and Propagation, IEEE, Berlin, 2009.
Communications, IEEE, Chengdu, 2018, pp. 1309–1313.

258