Design of Wireless Sensor Network in the Railway

Nagateru Iwasawa, Tomoki Kawamura, Michiko Nozue, Satoko Ryuo and Nariya Iwaki

Signalling and Transport Information Technology Division, Railway Technical Research Institute, 2-8-38,

Hikari-cho, 185-8540, Kokubunji-shi, Tokyo, Japan

Keywords: Wireless Sensor Network, Railway, Condition Monitoring System, Wi-SUN, 920 MHz, Slopes.

Abstract: In recent years, research and development on the condition monitoring systems using the wireless sensor

network in the railway have been proceeded. However, there are few cases of the wireless sensor network

design based on the features of the railway environment. In this paper, we propose the procedure to design

the wireless sensor network in the railway. And we introduce the demonstration test and the result in the

railway slope based on this procedure.

1 INTRODUCTION

The railway is the essential transportation mode in

many countries. Especially, in Japan, passenger

transport in FY2015 stood at 24.3 billion persons,

and freight transportation stood at 43.21 million tons

(Ministry of Land, Infrastructure, Transport and

Tourism, 2016). However, many Japanese railway

infrastructures were built before the 1970's, so the

average age of many tunnels and bridges is over 60

years.

To maintain and manage these aged facilities

properly is important for safe and stable operation in

railways. Therefore, regular inspections are carried

out on the structures once every two years, the

soundness of the structures is evaluated according to

the inspection results, and, if necessary, repair and

replacement and so on are carried out (Railway

Technical Research Institute, 2007).

In recently, condition monitoring by the WSN

(Wireless Sensor Network) has attracted attention,

and research on the condition monitoring system

utilizing the WSN is under way in the railway as a

matter of course. By monitoring the status all the

time with the WSN, it is expected that we can take

necessary measures based on right timing. However,

there is concern that the railway environment may

not be so created that, in there, we can collect data

frequently not enough to monitor the structure states

unless network design is properly made, since in the

railway environment, areas with different radio

environment such as urban areas and mountainous

areas are mixed and there are many metal objects.

Therefore, we propose a procedure for designing

a WSN in railway environment considering these

problems. Furthermore, based on the proposed WSN

design procedure, a WSN using the 920 MHz band

wireless communication standard Wi-SUN

(Wireless Smart Utility Network) was

experimentally introduced to an actual railway slope

and a demonstration test was conducted. In this

paper, we report the outline and results of the

verification test.

2 CONDITION MONITORING

USING THE WSN IN THE

RAILWAY

2.1 Condition Monitoring System

In the condition monitoring system using the WSN

that we are working on in this research, data is

collected by the WSN installed in the object to be

monitored, and these data are transmitted to the

M2M cloud via the Internet network and

accumulated (Fig. 1). These data can be viewed or

downloaded via the Internet network as necessary.

2.2 WSN

The WSN handled in this paper consists of a

gateway, relays, and wireless sensors. The wireless

sensor consists of a sensor for measurement and a

wireless terminal, and is attached to the object to be

122

Iwasawa, N., Kawamura, T., Nozue, M., Ryuo, S. and Iwaki, N.

Design of Wireless Sensor Network in the Railway.

DOI: 10.5220/0006638101220127

In Proceedings of the 7th International Conference on Sensor Networks (SENSORNETS 2018), pages 122-127

ISBN: 978-989-758-284-4

monitored. The data measured by the wireless sensor

is wirelessly transmitted to the gateway. If they

cannot be directly transmitted from the wireless

sensor to the gateway, they are transmitted to the

gateway via relay by multi-hop transmission. And

the collected data is sent to the cloud through a

public or private network and accumulated. In this

paper, the gateways, the relays and the wireless

sensors are collectively called nodes.

Figure 1: An example of the condition monitoring system

using the WSN.

WSNs have static and dynamic networks

(Bakaraniya and Mehta, 2012; Potdar et al., 2009).

The static network is a network for making data

transmission by a preset route. On the other hand,

the dynamic network is a network that

communicates between nodes and autonomously

composes.

The static network has the advantage that the

power load at each node can be calculated in

advance because the route is designed beforehand.

Thus, it is possible to formulate the battery design of

each node and an efficient battery replacement plan.

However, when a route change occurs due to the

addition of a node, it is necessary to change the

setting for the nodes on the route. Also, when a node

failure occurs, sensor data from that node and sensor

data passing through that node cannot be obtained

until the node is exchanged.

On the other hand, since each node

autonomously makes a route in the dynamic

network, it is not necessary to manually change the

route due to the addition of a node. Unlike the static

network, it does not always use the same route, so it

is not possible to calculate the power load of each

node in advance. So, for such a dynamic network,

we developed a method of estimating the battery

load by stochastically changing the route in time

series by the Monte Carlo method (Kawamura et al.,

2016). The details are omitted in this paper.

2.3 Problems in Designing WSN in

Railway Environment

As the features of the railway environment, it can be

cited that a wide area which urban areas and

mountainous areas between which there may be a

large difference in the radio wave environment

extends linearly and long, and that there are plenty

of metallic objects that are liable to hinder radio

wave propagation. In addition, from the viewpoint of

safety and physical conditions, restrictions may be

placed on the installation location and installation

height of nodes. For these reasons, the

communicable distance varies depending on the

installation location, and there is a concern that

communication quality for collecting necessary data

cannot be secured after their installation. Therefore,

a method of designing a network in consideration of

the environment of the installation site was

examined.

3 PROCEDURE OF WSN DESIGN

There is research on the design of WSN, but there

seems not to be found the network design

considering the railway environment as mentioned

above (Hodge et al., 2015; Tiwari et al., 2007; Xu et

al., 2005; Youssef and El-Sheimy, 2007). So we

propose a design procedure for introducing the WSN

in railway environment. It is explained below.

STEP1 Determination of objects and items to be

monitored

The railway operator decides the objects and the

items to be monitored. And it sets the collection

frequency and acceptable arrival rate of the data

required according to the monitoring item. In the

following procedure, the network is designed so as

to satisfy the required specifications set here.

STEP2 Determining the location of the wireless

sensors and gateway

It selects the location of the wireless sensor

according to the monitoring items determined in

STEP1. In addition, the location of the gateway is

selected the area of public or private network, and a

fixed power source is available, if possible.

STEP3 Survey of installation environment

In order to determine the location of the relay, it

confirms the location where the relay cannot be

installed and check the line of sight and obstruction

between the wireless sensor and the gateway. In

accordance with the installation environment, we

Sensor Data Server

Application Server

：Wireless Sensor

：Gateway

Wireless Sensor Network

Cloud

Data Representation

and Download

Internet

Sensor Data Transmittion

Design of Wireless Sensor Network in the Railway

123

appropriately select the frequency band and the

communication standard to be used. In addition, it

conducts a radio wave environment survey and so on

to derive communicable distance in the installation

environment.

STEP4 Derivation of communicable distance in

installation environment

From the result of the survey of the radio wave

environment at the installation location implemented

in STEP3, it derives the communicable distance in

this environment. The term "communicable

distance" means the distance at which the data

arrival rate between the nodes which results in

achieving the data arrival rate of STEP1 is secured.

STEP5 Determination of the installation location of

relay

If data cannot be transmitted directly from the

wireless sensor to the gateway, multi-hop

transmission is performed. If the wireless sensor can

be transmitted via another wireless sensor between

the wireless sensor and the gateway, there is no need

to take additional measures. However, if the number

is insufficient, dedicated relays should be installed.

By setting up this relay, we attain the achievement

rate set in STEP1. The details of the method of

determining the installation location of the relay

apparatus will be described Chapter 4.

STEP6 Power supply design

Since wireless sensors and relays are not necessarily

installed in locations where fixed power supply is

available, and there is a possibility that many nodes

may be installed for one monitoring location,

basically, power supply by a battery or an energy

harvesting is provided. Therefore, in accordance

with the network operation period and so on, the

design of battery capacity and energy harvesting

generation capacity and so on are properly

determined.

4 THE DETERMINATION

METHOD OF RELAY

LOCATION

We describe the method of determining the location

of the relay in STEP5 of chapter 3. This method

minimize the number of relays and satisfies

constraints given such as the specifications of the

nodes and the position of the obstacles. Also, since

various shielding objects are present along the

railway track, it is necessary to consider the

influence on the wireless communication by the

shielding object in determining the placement of the

relays. It is difficult to consider the influence of the

shielding objects in the conventional relay

placement determination method. In this paper, we

propose a method to decide the effective placement

of the relays of WSN, taking into consideration the

influence of shielding objects using mathematical

optimization. The objective function of this method

can be formulated as the minimization of the number

of relays as shown in Equation (1).

[Objective function]

min(R

num

)

(1)

[Constraints]

i,gat

= 1

(2)

(x,y) ≠ N

(x,y)

(3)

In the above equations, R

num

is the number of relays,

i,j

is the reachability matrix, r

i,j

＝1 if there is a route

by which data can reach node j from node i, and r

i,j

＝0 if there is no reachable route. Also, P

(x,y) is a

position (x coordinate, y coordinate) of the relay

device i, and N

(x,y) is the position (x coordinate, y

coordinate) where the relay cannot be installed.

Equation (2) represents the constraint relating to the

arrival of data from each wireless sensor at the

gateway, and r

i,gat

represents the reachability of the

gateway from the wireless sensor i by data. Equation

(3) represents the constraint relating to the position

of the relay.

Here, the position N

(x,y) where the relay cannot

be installed included in the constraint condition is

given as input, and it shall be set according to the

conditions of the environment where the WSN is

installed. In addition, the reachability matrix r

i,j

calculated by the following procedure by giving as

input such conditions as the position of the gateway,

the number of wireless sensors, the position of each

wireless sensor, the communication distance of the

wireless devices, and the position of the

obstructions.

STEP1 Generation of the adjacency matrix based on

the communication distance

STEP2 Updating the adjacency matrix based on

internode of line of sight

STEP3 Calculation of the reachability matrix based

on the adjacency matrix

Details of the above procedure are shown below.

SENSORNETS 2018 - 7th International Conference on Sensor Networks

124

4.1 Generation of the Adjacency

Matrix based on the

Communication Distance

In this paper, we consider the reachability matrix

showing the reachability of one of the nodes from

another by data using the adjacency matrix in the

graph theory. The adjacency matrix expresses the

presence or absence of the relationship between

nodes in the graph, and the adjacency matrix of the

graph consisting of n nodes is an n × n square

matrix.

Here, if the adjacency matrix is a

i,j

if there is an edge from node i to node j, a

i,j

= 1.

if there is no edge from node i to node j, a

i,j

= 0.

In this paper, the gateway, the wireless sensors, and

the relays are assumed to be the nodes in the

adjacency matrix, and the availability of

communication between each node is expressed as

an edge. That is, a

i,j

= 1 when communication from

node i to node j is possible, and a

i,j

= 0 when

communication from node i to node j is impossible.

Here, the determination of whether or not

communication is possible between the nodes is

made as follows using the communication distance

of the wireless devices of the wireless sensor or

relay given as the input condition.

if D

i,j

<= C

: Communication is possible (a

i,j

= 1),

if D

i,j

> C

: Communication is impossible (a

i,j

= 0)

Where, D

i,j

is the distance between nodes, C

is the

communication distance of each wireless device.

By performing the above judgment between any

pair of all the nodes, the adjacency matrix is

generated here.

4.2 Updating the Adjacency Matrix

based on Internode of Line of Sight

Here, the adjacency matrix generated in (1) is

updated based on the presence or absence of the

internode of line of sight. The presence or absence of

the internode of line of sight is determined based on

the position of the obstructions given as input. As

shown in Fig. 2, the position of the obstructions is

input as the gateways of a line segment constituting

the area where the obstructions exist like Li (x1, y1,

x2, y2). In this paper, the presence or absence of the

internode of line of sight is judged by the possibility

of intersection of a line segment constituting a

certain area of the obstructions and a line segment

connecting the nodes. Here, assuming that the two

line segments are L1 (x1, y1, x2, y2) and L2 (x3, y3,

x4, y4), the two line segments intersect when the

following equation (4) is satisfied.

tc × td < 0

(4)

Where,

tc = (x

- x

)(y

- y

)+(y

- y

)(x

- x

td = (x

- x

)(y

- y

)+(y

- y

)(x

- x

Figure 2: The coordinates of the line segment.

Here, the intersection determination is made

based on the above equation (4), and if any of the

line segments intersect each other as a result of the

judgment, it is determined that there is non-line of

sight and the adjacency matrix is updated as a

i,j

= 0

(communication is impossible).

4.3 Calculation of the Reachability

Matrix based on the Adjacency

Matrix

Here, the reachability matrix is calculated based on

the adjacency matrix calculated above. The

reachability matrix can be calculated by the

following procedure.

STEP1 Add unit matrix I to adjacency matrix A

STEP2 Under the Boolean algebra operation, A + I

is repeatedly multiplied by r times until the state

represented by the following expression (5) is

obtained

(A+I)

r-1

≠ (A+I)

= (A+I)

r+1

(5)

(A+I)

r+1

obtained by the above calculation is a

reachability matrix. In this way, in the method

proposed, the reachability matrix is calculated based

on the communication distance of the wireless

devices and the line of sight between the nodes.

5 THE DEMONSTRATION TEST

ON RAILWAY SLOPE

We conducted a demonstration test on the railway

obstruction

(x1,y1) (x2,y2)

(x3,y3)

(x4,y4)

L1(x1,y1,x2,y2)

Design of Wireless Sensor Network in the Railway

125

1cell = 10m × 10m, N: the relays installation impossible, O: the obstructions,

S: the wireless sensors’ location

Figure 3: The result of survey of the installation environment.

1cell = 10m × 10m, N: the relays installation impossible, O: the obstructions,

S: the wireless sensors’ location, R: the relays’ location

Figure 4: The result of the method to determine the location of relays.

slope, in accordance with the procedure proposed in

chapter 3.

5.1 Wi-SUN

In Japan, the 920 MHz band is allocated as the ISM

band in July 2012, and the application of the band to

the WSN is progressing. Along with that, the

development of the 920 MHz band LPWA (Low

Power Wide Area) wireless module is progressing.

The LPWA includes the LoraWAN, SIGFOX, Wi-

SUN, and the like. Compared to other LPWAs in the

same 920 MHz band, the Wi-SUN has the advantage

that, although the communication distance is inferior,

the transmission speed is 200 kbps and multi-hop

transmission is possible (Harada et al., 2017).

Therefore, it can be said that it is suitable for a WSN

which requires high scalability due to such changes

as the addition of wireless sensors, or one to be

introduced sequentially. So we decided to design the

WSN for monitoring the railway slopes by using

Wi-SUN.

5.2 WSN Design on Railway Slope

First, we decided the location of the wireless sensors

on the assumption of detecting the sign of land slide

and land collapse by measuring the inclination and

the soil moisture on the slope. And we decided to

install the gateway in the location that can supply

fixed power.

Next, we conducted survey of the installation

environment, and confirmed the location where the

relay can be installed and the shielding object and so

on from topographic conditions based on

topographic map (Fig.3). Then, we applied the

proposed method in Chapter 4 to determine the

setting location of relays for the wireless sensors far

from the gateway in Fig. 3. Fig. 4 shows the location

of the relays by the proposed method. Incidentally,

the wireless sensors and relays were powered by

solar panels in this demonstration test.

5.3 The Result of the Test

We installed 5 inclination wireless sensors, 2 soil

moisture wireless sensor and 3 relays. Fig.5 shows

the state of installation the nodes on the slope.

We accumulated the sensor data by the designed

network for about 3 months. As a result, except for

some sensors, the sensor data arrival rate was more

than 99%. But some of those sensors have lost data

from one day. We know that rain and snowfall don’t

have a big influence on radio wave propagation in

the 920 MHz (ITU-R, 1998; Iwasawa et al., 2016).

One of the reasons for this may be that the plants

grew higher than the antenna of the wireless sensors

and the relays. In fact, we saw the images from the

camera on the site and confirmed that the wireless

sensors and the relays were buried with plants. Also,

we confirmed that radio waves attenuate when there

are plants between the nodes. In the future, we think

that it is necessary to design the WSN considering

such attenuation.

6 CONCLUSIONS

In this paper, we introduced the procedure for

designing the WSN in the railway environment. And

we reported the demonstration test to the railway

slope based on the procedure. As a result, we found

that the WSN design considering the change of the

N N N N O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O

S N N N N N N N N O O O O O O O O O O O O O O O O O O N N N N N

N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N

C N N N N N N N N N N N N N N N N N N N

O O O O O O O O S

O O O O O O O O

R N N N N O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O

S N N N N N N N N O O O O O O O O O O O O O O O O O O N N N R N N

N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N

C N N N N N N N N N N N N N N N N N N N

O O O O O O O O S

O O R O O O O O O

SENSORNETS 2018 - 7th International Conference on Sensor Networks

126

Figure 5: The state of installation the nodes on the slope.

surrounding environment is necessary. So we will

improve the procedure by organizing the idea of

margin due to environment change in the future.

And we will plan to improve the proposed method

so that we can consider the dynamic environment

such as train movement.

ACKNOWLEDGEMENTS

The research results have been achieved by

"Research and Development on Fundamental and

Utilization Technologies for Social Big Data," the

Commissioned Research of National Institute of

Information and Communications Technology

(NICT), Japan.

REFERENCES

Bakaraniya, P., Mehta, S., 2012. Features of WSN and

Various Routing Techniques for WSN: A Survey, In

International Journal of Research in Engineering and

Technology.

Harada, H., Mizutani, K., Fujiwara, J., Mochizuki, K.,

Obata, K., Okumura, R., 2017. IEEE 802.15.4g Based

Wi-SUN Communication Systems. Transactions on

Communications. IEICE.

Hodge, V. J., O’Keefe, S., Weeks, M., Moulds, A., 2015.

Wireless Sensor Networks for Condition Monitoring

in the Railway Industry: A Survey, In Transactions on

Intelligent Transportation Systems. IEEE.

Iwasawa, N., Ryuo, S., Kawamura, T., Kawasaki, K.,

Nozue, M., 2016. Transmission Performance

evaluation of Wi-SUN Wireless Sensor Network

During Snowing, In 23

J-RAIL’16. IEEJ.

Kawamura, T., Ryuo, S., Iwasawa, N., 2016. Power

Consumption Prediction Method for Train-Health

Monitoring Wireless Sensor Networks, In 23

J-

RAIL’16. IEEJ.

Ministry of Land, Infrastructure, Transport and Tourism,

2016. Annual report on railway transportation, MLIT,

Tokyo.

Potdar, V., Sharif, A., Chang, E., 2009. Wireless Sensor

Networks: A Survey. In AINA’09, 23

International

Conference on Advanced Information Networking and

Applications Workshops. IEEE computer society.

Railway Technical Research Institute, 2007. Railway

structures maintenance standard and commentary

[earth structure] (embankment and excavation),

Maruzen, Tokyo.

ITU-R, 1998. Specific Attenuation Model for Rain for Use

in Prediction Methods, In ITU-R Recommendations,

ITU, Geneva.

Tiwari, A., Ballal, P., Lewis, L. F., 2007. Energy-Efficient

Wireless Sensor Network Design and Implementation

for Condition-Based Maintenance, In Transactions on

Sensor Networks. ACM.

Xu, K., Wang, Q., Hassanein, H., Takahara, G., 2005.

Optimal Wireless Sensor Networks (WSNs)

Deployment: Minimum Cost with Lifetime Constraint.

In WiMob’05, Wireless and Mobile Computing,

Networking and Communications. IEEE.

Youssef, M., El-Sheimy, N., 2007. Wireless Sensor

Network: Research vs. Reality Design and

Deployment Issues. In CNSR’07, 5

Annual

Conference on Communication Networks and Services

Research. IEEE computer society.

Gateway

Inclination wireless sensor

About 400m

About 100m

Relay

Soil moisture wireless sensor

Inclination wireless sensor

Design of Wireless Sensor Network in the Railway

127