Using Geofencing for a Disaster Information System
Gondil, Guide ,IEEE, Saraswati Yadav, Member
,IEEE, Komal Wagh, Member, IEEE,
Shreya Thakur, Member, IEEE
Abstract—This paper proposes a disaster
information system with the geofencing equipment
the association of users as well as provide in a row of the risk in favour of
them. The system is collected of client-server planning; the head waiter collect
danger information from a variety of information sources and the client watches
the user on the way to inform the in sequence as the need arise. To notice the
user’s society the client creates a virtual barrier called geofencing at the
dangerous region based lying on the danger information stored into the head
waiter, and monitor the user’s entrance as well as outlet of the fence. so the scheme
can deliver warning and advices suitable to specific users failing. We
implemented a prototype system and evaluated the accuracy of the system. The
location of the user was detected among high correctness while incoming the
fence, but the correctness was low when exiting the fence.
location-based services; navigate user swift; iOS application
Japan suffers big damage from natural disasters
every year. The cause of this is due to no correct information to the people
who need it. There is a report entitled “Evacuation instructions and
questionnaire survey about evacuation directive” by the Japanese Cabinet Office
1. Table I and Table II show the questionnaire results. In Table I, they
asked the behaviors when people knew evacuation instructions or evacuation
directives. The answer “Remained in their houses” was the first place. In Table
II, top reasons for this were “They thought evacuation was dangerous because of
heavy rain during the night” and “They did not think that they suffer from the
disaster”. Thus, it is obvious that the current information delivery method is
not suitable to residents. In particular, the current evacuation advices and
instructions do not inform risk enough, because the scope of these advices and
instructions are too wide.
If a system can deliver directly
such risk information only to people who need it, the damage may be possibly
reduced. This research aims at developing a system that detects people’s
movement and delivers risk information. For this purpose, we inspected the
accuracy of detection of people’s movement using geofencing, which dynamically
defines geographic area of interest. By using geofencing, it is possible to
detect entries and exits of people at the specific area. Thus our system can
deliver what is happening at a particular area directly to the users.
TABLE I. THE BEHAVIOR WHEN PEOPLE KNEW EVACUATION INSTRUCTIONS
OR EVACUATION DIRECTIVES (KANI CITY) (THE TOP FOUR ITEMS)
TABLE II. THE REASON WHY PEOPLE REMAINED IN THEIR HOUSE INSTRUCTIONS
OR EVACUATION DIRECTIVES (KANI CITY) (THE TOP FOUR ITEMS)
Our system delivers risk
information timely to specific users who are in the area where a disaster has
occurred or may occur with high probability.
We assume that each user has a
smart phone with position detection and Internet connection capabilities.
Because the users usually handle their smart phones, they can also acquire
information smoothly when a disaster occurs. Moreover, it is possible to detect
the user’s current location and receive information on the disaster from the
978-1-5090-0806-3/16/$31.00 copyright 2016 IEEE
ICIS 2016, June 26-29, 2016, Okayama, Japan
THE PROPOSED METHOD
A. What is
Geofencing is a mechanism that
makes a virtual fence in a specific area 2. The application sets a geofence
at a dangerous area and gives risk information to the user. Fig. 1 shows the
movement against a geofence.
Geofencing action example
In order to define a fence, the
coordinate (latitude and longitude) of the place are required. A circular area
is defined by the coordinate and radius. A geofence is set to the circular area
B. How to
use the geofencing
The system using geofencing is
possible to deliver the disaster information to the user who has just entered
the fence. In this research, we implement geofencing with the Core Location
framework of iOS. This framework provides a detection of the entries and exits
of the user with the observation of a specific geographic region. The
geographic region is an area defined by a circle with a specified radius around
a known point on the earth. Every time the user crosses the boundary of the
region, the system generates an event for our application. This enables the
notification of the disaster information. That is, by using the observations of
geographical area, it is possible to detect user behavior in the same manner as
the definition of geofencing.
Moreover, the system does not
report the event until the user goes into the region further from the boundary
plus a system-defined cushion distance. This cushion value prevents the system
to generate numerous events while the user is traveling close to the boundary.
The cushion distance is determined by the hardware and the location
technologies that are currently available 4.The system navigate the user to
come out from disaster affected area.
IV. SYSTEM CONFIGURATION
The system is composed of
clients, a server and information sources. Fig. 2 shows the system structure.
Each client is an application
program running on iOS. It connects to the Internet and obtains the information
from the server. Moreover, it defines a geofence based on information from the
server, and notifies disaster information to the user. The client is
implemented by using Xcode7 and swift2, and tested by iOS simulator and real
The server is a web application
running on Linux (Centos7). It is composed of Apache, MariaDB, and PHP. The
server acquires disaster information from information sources. It analyzes the
information and stores the result in a database. The database is used to define
a fence by the client.
An information source is the RSS
file of Weather Warnings and Advisories that Yahoo! JAPAN provides 5. The RSS
file, provided in the RSS 2.0 format, contains “Special alert,” “Weather
Warnings,” or “Advisories” across Japan. The RSS file is updated regularly
according to the information announced by the Japan Meteorological Agency.
As an example, suppose that the
possibility of flood increased due to a heavy rain continued for long time. As
the result, a flood warning has been issued to the area.
Then, the server’s PHP program
acquires the warning by means of RSS files from the Internet. Then, it stores
the disaster information in the database. On the other hand, a client
periodically accesses the server to check new information. The server program
retrieves the database based on the client’s request and returns the result
including location data to define a fence in a JSON format. In this research,
we assume that the specification of the fence is decided on the server-side.
The client sets the fence by using the CLCircularRegion class.
Then, the client starts
monitoring of the entry and exit of the user to the fence by calling the
startMonitoringForRegion method of the CLLocationManager object. When the user
enters the fence, the locationManager:didEnterRegion method is invoked. Then,
the client warns the user that you have entered the dangerous area. When the
user exits the fence, the locationManager:didExitRegion method is invoked.
client notifies the user that you
have exited the dangerous area.
Fig. 3 shows a flowchart.
Work flow of the system
warning is cancelled, the client finds no warning on the server. Then, it calls
the stopRangingBeaconsInRegion method of the CLLocationManager object to stop
monitoring of the entry and exit of the user to the fence. Fig. 4 shows the
screen when operating in the foreground.
Screen shot at foreground
When operating in the background,
the notification is performed using the notification banner. In the background,
“Background fetch” of “Background Modes” is used to acquire disaster
information automatically. Background fetch enables the application to
regularly download and process a small amount of contents from the network.
Fig. 5 shows the screen when operating in the background.
Screen shot at background
of the experiment
experiment was conducted to confirm whether information delivery works actually
using geofencing. This subsection describes our experiment method. A user
carried a real iPhone device, and defined a fence. Then the user entered and
exited the fence. The fence was installed in a circle whose radius was 100m
centered on Tokyo Senju Campus of Tokyo Denki University. The fence size was
determined for small-scale disasters such as landslips.
The measuring range was 300m from the center
of the fence. We recorded the behavior of the application while moving every
10m distance. The measurement was done three times with Wi-Fi on and off.
of the experiment
Table III shows the experiment
result, and Fig. 6 shows the place where notice movement was done. The “NA” in
that the system did not detect the entry or exit of the fence, and then the
user could not receive risk information.
III. RESULT OF THE EXPERIMENT
Location where notice of movement was done
In the case that of Wi-Fi was on,
the user got information at the entry or exit of the fence. However, the
location was not accurate. When entering the fence, the user was notified at
the location of 120-130m, while the radius of the fence was 100m. That is, the
risk information was delivered to the user at the location of 20m-30m outside
of the fence.
When exiting the fence, the user
was notified at the location of 220-230m. That is, the notification of the exit
was at a distance of more than 100m from the fence.
On the other hand, in the case
that of Wi- Fi was off, the user did not get any information for both entry and
exit of the fence.
When entering the fence with
Wi-Fi on, risk information was delivered to the user at 20-30m front of the
fence. It is considered that the error occurred due to the buffer region of the
fence and the positioning error of the GPS. However, the notification was
performed at a position before entering the fence, it is possible to provide
the information before the user enters a dangerous area.
When exiting the fence with Wi-Fi
on, the notification was performed at the location more than 100m away from the
fence. We consider the reason of the error for exit is different from that for
the entry. More study is necessary.
The system could not detect the
entry and exit of the fence in the case that of Wi-Fi is off. This is because
the location is computed more precisely using information from the Wi-Fi by the
location information service provided for iOS devices.
VII. RELATED WORK
The term “geofencing” is used
from around 2000. It appeared in research literature by Munson and Gupta 6 in
2002. Geofencing is one of core technologies today for location-based services
(LBS) including advertising, tracking, and risk management. Szczytowski 7
proposed an approach based on combining geofencing with social networking
systems (SNS) to organize unstructured information collected from SNS. Yelne
and Kapade 8 designed a help-me application running on an android operating
system based on geofencing. Detection accuracy and power consumption are very
important for geofencing applications. Nakagawa et al. 9 proposed a method
for position detection whose activation frequency is determined by speed toward
the target spot. Alsaqer et al. 10 investigated accuracy and battery-use of
Esri’s geo-trigger service in small, outdoor, geo-fenced areas.
A system to present disaster
information based on person’s movement was proposed. We implemented an
experimental system by using geofencing and evaluated the system in an urban
area. We confirmed that our system notifies disaster information when a user
enters the fence with Wi-Fi on by the experiment. The location was at 20-30m
outside the fence. When exiting the fence with Wi-Fi off, we found that the
information is delivered at the place more than 100m outside the fence. Wi-Fi
is necessary for precise detection of location by using geofencing.
large-scale disasters, the fence will be several kilometers of length. Further
study is necessary to evaluate the system in case of larger fence sizes. Improvement
of the location accuracy is also very important to deliver risk information
timely to users. Our system should be able to define multiple fences at the
same time to support real natural disasters. Information sources also should be
added to our system, including government agency announcements and social
networking services.It also navigate the user.