Consumer Propensity and Location Analysis based Real-time
Location Tracing Advertisement Service Design and Implementation
Real-time Location based Advertisement System
Daehee Won, Yoonsoo Kim, Hangki Joh, Intae Ryoo
and Dougyoung Suh
Kyunghee University, Yong-in, 446-701, Republic of Korea
Keywords: Advertisement Service, Location based Service, Consumer Propensity, Real Time Location based Service.
Abstract: While distributing Android free of charge, Google intended to expose its advertisements on the platforms to
seize users eyes and make profits. However, smart phones are kept in bags or pockets during most of the
time instead of showing screens in front of users eyes. If the time during which users eyes cannot be
seized becomes longer advertisement effects will decrease as much. In this study, in order to solve these
problems, consumers movement paths are grasped using continuous screens based on the results of
analyses of consumer propensity to replay advertisement images. Advertisement image replay lists are
composed of related advertisements based on the key words set by consumers. The relevant project was
named as shADow meaning Advertisements that follow like shadows.
1 INTRODUCTION
In the revolution of smart phones headed by iPhone,
Location Based services (LBS) are in a trend of
being expanded from existing technologies that use
the GPS (Global Positioning System) or Cellular
Networks to the WPS (Wi-Fi Positioning System), a
positioning technology that uses wireless Aps
(Access Point) (Robert, 2009) and diverse location
based services utilizing the WPS are being provided.
(Kubber, 2005); (KAIT, 2010)
From the beginning of mobile phone supply,
location based services have been receiving attention
as one of areas of high growth potential since this is
an area that is closest to mobility which is one of the
advantages of mobile phones. However, Feature
Phones failed to bring about the activation of
location based services due to problems such as high
data communication costs, lack of contents and the
issue of the openness of platforms.
The popularization of smart phones brought
about great changes in location based services and
thus location based services are rapidly rising to
become core services of mobile communications. It
has become possible to develop diverse location
based applications through smart phone platforms
such as Android and iPhone. To this end, not only
the existing positioning technologies through GPS
but also indoor/outdoor wireless positioning methods
and technologies through diverse data are necessary.
The WPS which is one of the aforementioned
technologies is a positioning method using Wi-Fi,
that is, wireless APs of WLAN networks. In general,
terminals find positions by measuring the intensities
of signals (RSSI) from wireless APs and calculating
signal transmitting distances based on signal
attenuation. The WPS method is frequently used
because it can provide high-speed Internet services
to many users in regions where there are many
indoor movements and it can be conveniently
installed and managed. Positioning methods utilizing
Wi-Fi can be largely divided into three types;
Fingerprinting, Cell-ID and triangular surveying
(Kim, 2006).
Figure 1: Conceptual diagram of shadow.
171
Won D., Kim Y., Joh H., Ryoo I. and Suh D..
Consumer Propensity and Location Analysis based Real-time Location Tracing Advertisement Service Design and Implementation - Real-time Location
based Advertisement System.
DOI: 10.5220/0004063201710179
In Proceedings of the International Conference on Signal Processing and Multimedia Applications and Wireless Information Networks and Systems
(SIGMAP-2012), pages 171-179
ISBN: 978-989-8565-25-9
Copyright
c
2012 SCITEPRESS (Science and Technology Publications, Lda.)
Initial applications using smart phones location
based services were mainly concentrated on services
such as Navigation, positioning, security, etc.
Currently, those applications are provided with
diverse forms of services such as user searches,
place searches, augmented reality and SNS. As
revealed in the recent mobile advertisement business
M&A between Google and Apple, mobile
advertisements count as one of areas of the highest
possibility of growth in the market. In this study,
services were implemented that would always exist
in front of consumers eyes through real time
positioning using smart phones to enhance
advertisement effects. These services are called
shadow (Shadow + Advertisement) in this paper.
2 RELATED STUDIES
2.1 Indoor Positioning Technology
2.1.1 Fingerprint based Positioning
Technique
The fingerprint based indoor positioning technology
is one of most frequently studied techniques since it
was applied to Wi-Fi in the early 2000s. (Bahl,
2000) When entering into an environment where an
AP has been installed, this is used to measure the
intensity of signals from the AP and compare the
result with the signal intensities of the RPs
(Reference point) stored in the DB in advance to
assume that the RP that has the most similar
characteristics as the location of the terminal.
This technique is divided into a training stage to
grasp the characteristics of signals in preset areas
and store the characteristics in the DB and a
positioning stage to determine the location with the
characteristics. Although this technique has
disadvantages in that it requires preceding work to
set areas in advance before positioning, grasp the
characteristics of signals by area and store the results
in the DB and that these processes should be
repeated every time the environments in positioning
areas are changed, it has advantages in that it is not
restricted very much by surrounding environments
since it reflects the information on noises and
environments in surroundings on positioning and
that the accuracy of positions determined through it
is excellent (Cho, 2007).
2.1.2 Cell-ID
The Cell-ID based positioning technology known as
Cell-tower based mobile phone positioning
technology refers to a technology to estimate the
present position based on the identification numbers
of Wi-Fi APs in the surroundings.
If a position where an APs has been installed in a
certain building is stored in the DB, the
identification number of the AP accessed (or
searched) by the terminal will be searched from the
DB to assume it as the present position of the
terminal. Unlike cell-towers of which the
transmission distances reach several hundred meters
to several kilo meters, each Wi-Fi AP has short
transmission distances of around several ten meters
to 200m at the maximum depending on
environments. Since short transmission distances
mean high accuracy as much, this method is
meaningful in that it has relatively high accuracy
while being simple. This technique is mainly utilized
to judge whether terminals exist in the relevant area
and its accuracy is enhanced in proportion to the
density of installation of Wi-Fi APs at certain
intervals.
2.1.3 Triangular Surveying
The triangular surveying is a technique frequently
used during positioning using not only wireless APs
but also satellites or base stations. The RSSIs
measured from at least three points of which the
coordinates are known are converted into distances.
The measured distances are used as radiuses to draw
circles at individual points and the area where these
three circles overlap each other is estimated to be the
present position.
RSSIs are converted into distances using the
following Friis formula.



(1)
where, c indicates the propagation velocity, f
indicates the frequency and L indicates signal
transmission loss. By substituting the distance d
obtained from each point into the formula for circles
as shown under (2), the position to be measured can
be obtained in the method as shown in Figure2.
 
  
 
   
 
   
(2)
However, this method is quite vulnerable to noise
that frequently occur in radio waves since resultant
distance values exponentially increase as RSSIs
become lower. Therefore, this method requires
studies to predict and respond to noises.
SIGMAP 2012 - International Conference on Signal Processing and Multimedia Applications
172
d
1
d
2
d
3
(x
1
, y
1
)
(x
2
, y
2
)
(x
3
, y
3
)
(x, y)
Figure 2: Triangular surveying using the Friis formula.
2.2 Examples of Indoor Positioning
2.2.1 RADAR (Bahl, 2000)
The RADAR introduced by Microsoft Research
Group in 2000 is the first Wi-Fi signal intensity
based indoor positioning system in the world. Before
positioning, the strength of signals (SS) transmitted
from the base station by reference point and the
signal to noise ratios (SNR) are measured and stored
in the DB through an off-line phase. Then, this
information is used to analyze the signals actually
received by the user in the Real-Time phase to
estimate the position. Despite that the RADAR is an
outcome of the Wi-Fi based indoor positioning
technology studied in the early stage, it uses the Wi-
Fi AP infrastructures installed in buildings as they
are and combines the DB measured in advance and
the characteristics of the signals received by the user
in real time to provide high accuracy.
2.2.2 Ekahau (Paul, 2009) (Ekahau)
The Ekahau has been developed to estimate the
positions of tags installed with Wi-Fi in hospitals,
marts, distribution facilities, factories, public
facilities, etc. and it uses already installed APs for
positioning using the Fingerprint method. This
solution is composed of Positioning Engines,
Managers, Applications (Finder, Tracker, Logger),
and Tags and the Tags are provided in diverse forms
depending on the purposes of utilization. Since the
Fingerprint method is used, a stage of collecting
fingerprints in the relevant area should be gone
through and the accuracy of positions is proportional
to the density of the collected fingerprints.
2.2.3 Playmap (Playmap)
The Playmap is an LBS+SNS form service published
by Hyundai MN Soft in 2009 and is an LBS
application that received a prize in the Mobile
Technology Grand Prize hosted by Korea
Communications Commission in 2009.
With wire/wireless interlocking place sharing
services consisting of the Playmap website and
applications, this has functions to search position,
explore routes and share information and provides
diverse functions combined with maps.
This enables sharing restaurants that provide
delicious foods encountered by users in their life,
places good for taking images and enjoyable travel
destinations and in particular, the post function and
the function to recommend popular places in real
time based on the degree of interest of users present
the direction of evolution of LBS to analyze users
use behavior by combining SNS elements with LBS
along with the business model of Foursquare.
2.2.4 Foursquare (Foursquare)
This is an LBS based SNS to enable users to inform
their locations and what they are doing and leave
memos through smart phones in order to share
information with friends.
If a Foursquare user press the check in button
when he has come to a certain place, he will validate
that he came to the place. As this check in value
increases, the user’s position will be enhanced and
the user can finally receive a badge.
As of April 2011, the number of Foursquare
users in the world exceeded 8 million. Using this
Foursquare, sales promotions of off-line shops are
conducted utilizing location based SNS and new
revenue models of LBS are being presented such as
the provision of gifts or discount services based on
the frequency of off-line visits.
3 REAL TIME POSITIONING
USING WIRELESS APS AND
ACCELERATION SENSORS
3.1 Removal of Noises in RSSI Values
RSSI measurement in smart phones that use Android
has an update period of 1500ms and a range of
around -20dBm ~ -100dBm. When pure RSSIs are
measured, noises may increase rapidly depending on
the performance of wireless APs and surrounding
Consumer Propensity and Location Analysis based Real-time Location Tracing Advertisement Service Design and
Implementation - Real-time Location based Advertisement System
173
environments. If these noises increase, RSSI values
will decrease rapidly. To respond to this, the
weighted averages (Pomalaza, 1984) as shown under
(3) are used.
(3)
In expression (3), the RSSI is the  measured
presently and
is the results of compensation for
noises. Expression (3) shows values that will make
RSSIs strongest against noises while being able to
immediately reflect measurement results determined
based on the results of many experiments. The
largest weighted values are given to RSSIs that have
been collected the most recently and adapted in real
time to the intensity of signals exchanged with
wireless APs to measure the RSSIs. The results of
experiments conducted using expression (3) are as
shown in Figure3.
Figure 3: RSSI measurement result.
Figure 4: Results of weighted average application.
Figure 2 shows the results of measurement of
RSSIs performed while changing the distance
between the wireless AP and the smart phone after
installing an N2emodel of ipTime at an end of 3m
wide 20m long corridor. It can be seen that the RSSI
became smaller as the distance between the wireless
AP and the smart phone increased while becoming
larger as the distance between the wireless AP and
the smart phone decreased. Figure2 shows the results
of measurement and there are many difficulties as
many noises are mixed into signals. On the other
hand, Figure 3 shows the results of RSSI
measurement corrected using expression (3). It can
be seen that noises decreased remarkably compared
to pure results of RSSI measurement.
3.2 Measurement of RSSI and
Acceleration Sensor Values
Wireless APs were installed at both ends of an actual
corridor and the values of RSSIs and acceleration
sensors applied with expression (3) were measured
for three cases; a case where the experimental
subject moved carrying a smart phone in his hand, a
case where the experimental subject moved carrying
a smart phone in his trouser pocket and a case where
the experimental subject moved carrying a smart
phone in his bag pocket. In all these cases, the values
were measured four times while the subject was
moving between the wireless APs and the two ends
of the corridors back and forth. The first two
measurements were done while the subject was
walking and the last two measurements was done
while the subject was running.
The upper parts of Figure 5~7 show the largest
values among measured values of the acceleration
sensors at the X, Y, and Z axes in lines and the lower
parts show the RSSIs of the two wireless APs as bent
line graphs of solid lines and dotted lines
respectively.
Figure 5: The values of the RSSIs of the two wireless APs
and acceleration sensors when the subject was walking and
then running carrying a smart phone in his hand.
Although the RSSI values were compensated for
SIGMAP 2012 - International Conference on Signal Processing and Multimedia Applications
174
noises using weighted averages, it is difficult to
determine locations with only RSSI values. However,
it is possible to judge whether the user switched the
direction using the RSSI values. This paper proposes
this in 3.3 Direction switching using RSSIs.
Figure 6: The values of the RSSIs of the two wireless APs
and acceleration sensors when the subject was walking and
then running carrying a smart phone in his trouser pocket.
Figure 7: The values of the RSSIs of the two wireless APs
and acceleration sensors when the subject was walking and
then running carrying a smart phone in his bag.
Although acceleration sensor values showed
differences among locations where the smart phone
was kept, these values showed a tendency that
measured values increased at once when the subject
started running after walking. Based on this, cases of
walking or cases of running can be judged on the
basis of a certain value. (Huang, 2010) This paper
proposes this in 3.4 Measurement of moving speeds
using weighted averages.
3.3 Direction Changes using RSSIs
With the results of removing noises from the RSSIs
measured from two wireless APs, a numerical
formula to predict direction changes can be made
like expression (4) using expression (3).







(4)
In expression (4),
values to check
whether the direction has been changed and

are the results of noise removal from
signals received from individual wireless APs using
expression (3). The weighted values here were
determined among values that could immediately
reflect measurement results while being the strongest
against noises based on the results of many
experiments. The largest weighted value is given to
the RSSI that has been the most recently corrected to
adapt to the intensity of signals exchanged with the
wireless AP in real time to measure the RSSI.
Whether the direction has been changed can be
determined through changes in the values of
in
expression (4). If
values decrease after showing
an increasing trend or increase after showing a
decreasing trend, it can be seen that the direction has
been changed. An algorithm of this is as shown in
Figure 8.


















Figure 8: Algorithm of direction changes by RSSIs.
In Figure 8, 
is the result of
expression (4) and  is the present state
of changes in RSSIs. The moment of a change in
 can be determined as the moment
where the direction changes. The results of
experiments of this are as shown in Figure 9.
In Figure 9, the bent line graph in a dotted line
shows the results of expression (4) and the bar
graphs in solid lines show time points at which the
direction was actually changed. It can be seen that an
infection point appears when around 2~3 RSSIs have
been measured and calculated after an actual
direction change. An Android smart phone has RSSI
renewal periods of 1500ms and perceives the
Consumer Propensity and Location Analysis based Real-time Location Tracing Advertisement Service Design and
Implementation - Real-time Location based Advertisement System
175
direction change within around 3000~4500ms after
an actual direction change.
Figure 9: Results of judgment whether the direction has
been changed using expression (4).
3.4 Moving Speed Prediction using
Acceleration Sensors
Though Figure 5~7 and paper (Huang, 2010), it can
be seen that acceleration sensor values suddenly
increase when the user starts brisk walking.
Acceleration sensor values are separately measured
for X, Y and Z axes and to compared on the basis of
the axis that has the largest value at each moment,
the largest value in the case of normal walking is
6.361706 obtained when the subject was normal
walking carrying the smart phone in his trouser
pocket and the smallest value is 2.12406525
obtained when the smart phone was put into the bag.
On the other hand, the smallest value in the case of
brisk walking is 8.119599 obtained when the smart
phone was put into the bag. Error ranges mat be set
and cases where the value is 8 or larger may be
regarded as cases of brisk walking, cases where the
value is 1 or larger but smaller than 8 may be
regarded as cases of walking and cases where the
value is smaller than is may be regarded as cases of
no movement. Numerical formulas for these cases
are as shown in Table 1.
Table 1: States of movements based on accelerometer
measured values.
Condition
State




Brisk walking




Normal walking




Stationary
Therefore, the states of users movements can be
known through the accelerometer. Table 2. Counted
versus measured steps at normal and walking brisk
speeds; effects of age” in (Melanson, 2004) shows
moving averages of moving speeds by the age of
users. It can be seen that whereas general adults walk
slower than 3MPH at normal speeds, they move
faster than 3.5MPH when they walk at high speeds.
Figure 10: Composition of advertisement lists through
consumers SNS.
4 SELECTION OF
ADVERTISEMENTS OF
INTEREST USING
CONSUMERS SNS
To expose advertisements that coincide with the
matters of consumers interest instead of non-
differential exposure of advertisements,
advertisements include key word information in
addition to video information. Key words comprise
words that are directly related with products under
the relevant advertisement. To collect these key
words, related key words are collected first using the
Open Mind Common Sense (Open Mind Common
Sense) which was a project of the MIT media
research room.
Thereafter, when the users smart phone accesses
the server and transmits its SNS ID as shown in
Figure 8, the server receives the most recent time
line from the SNS based on the SNS ID. The
propensity analysis algorithm searches the key words
in the server from sentences other than negative
statements from the most recent time line in
sequence. If a time line has the key word, the images
of the advertisement having the relevant key words
will be added to the list.
When the composition of advertisement lists has
been completed, the location of the video player is
changed based on the information the location of the
SIGMAP 2012 - International Conference on Signal Processing and Multimedia Applications
176
Figure 11: shADow Sequence diagram.
user and the advertisement images in the
advertisement lists are replayed in sequence. If
information on desired products appears, the user
may request the information through his smart phone.
At the moment the request has been received, the
server sends the price information, location
information, discount information, image
information, etc. for the product in the advertisement
to provide the information to the user.
5 RESULTS OF
IMPLEMENTATION AND
ANALYSIS
As proposed in chapters 3 and 4, shadow is executed
by the operation algorithm shown in Figure 9.
1. When requested by the advertiser, the
advertisement is registered in the shADow system.
Information such as videos, images and key words is
stored in the DB.
2. While installing shADow applications in his
smart phone, the user enters his SNS ID.
*
*
*
*
shADow system SNS Openmind.media.mit.edu
1. apply advertisement(Video, Image, Keyword)
2. join(Registering SNS ID, installing application)
Advertising agent
User
3. System visit around
4. transmit user's SNS ID
5. URL create using the SNS ID
6. Request the recently time line
7. Recently time line
8. Parsing, delete the negative string and rising period the positive string
9. Create URL using the applied keyword
10. URL page
11. Keyword parse
12. Matching to the advertisement keyword /
13. Advertisement play
14. Extraction RSSI, recognization to change the direction(remove noise using the weighted average)
15. Recognization user's moving speed using accelometer sensor
16. Send the recognization information
17. Change the location of video player
18. Confirmation
make the interested advertisement list
Consumer Propensity and Location Analysis based Real-time Location Tracing Advertisement Service Design and
Implementation - Real-time Location based Advertisement System
177
3. Using the Notification function of Android, the
user checks whenever a new wireless AP is found
whether registered wireless APs are in the vicinity.
4. The stored SNS ID is sent to the server so that
the server can compose advertisement image lists.
5~8. Using the SNS ID received from the user’s
smart phone, the server prepares URLs for receiving
recent writings to receive pages containing recent
writings and parse the pages. In individual writings,
negative statements are removed and the priorities of
positive statements are enhanced.
8~11. Using the registered key words, the server
creates URLs that can be used in Open Mind
Common Sense (Open Mind Common Sense) and
collects related key words.
12. The server checks the statements if they have the
key words collected from recent writings in SNS and
add the advertisement images of the relevant key
words to the list.
13. While replaying the files in the image list, the
server changes the position of the player based on
the location information received from the user’s
smart phone. The server was implemented as a
multi-server so that information from many clients
can be received simultaneously.
14. The Client was made to apply expressions (3)
and (4) to measured RSSIs to perceive direction
changes through Figure 8 and the measured values of
the acceleration sensor trough tables 1 and 2 and
send the speeds to the server every 300ms. The
screen used is as shown in Figure 11.
Shadow actually operates as shown in Figures 12~16.
Figure 10 is a scene where two beam projects were
connected to the server to illuminate the
experimental subject. The experimental subject
carries a smart phone in his hand and moves from
the left side to the right side and returns to the left
side without any operation of the smart phone. One
each wireless AP is installed on the left end and the
right end and based on the propensity analysis
through the experimental subjects SNS, a ramen
advertisement video is being replayed.
Through the direction change algorithm set forth
in Figure 8 and expression (2), changes in the
direction between the APs could be identified in
relation to changes in smart phone locations.
Through the experiment, it could be seen that a
direction change could be perceived only when data
for 3 to 4.5 seconds have been gathered and thus as
much delays occurred in direction changes.
Although direction changes were not completely
perceived since noises were not completely removed,
much more precise positioning was possible than
finding locations through triangular surveying.
Figure 12: Smart phone execution screen.
Figure 13: Connection of the shADow with the Client and
the start of the video player.
Figure 14: The start of the users movement.
Figure 15: Perception of user direction change.
Movement speed measurement was almost
accurate. The algorithm operated normally
SIGMAP 2012 - International Conference on Signal Processing and Multimedia Applications
178
regardless of locations where the smart phone was
kept or movement speeds. In the case of acceleration
sensors, no delay occurred because collection
periods were very short.
Diverse key words related with those included in
advertisements through Open Mind could be
compared with the key words of users interest.
Figure 16: User movement.
Figure 17: User movement.
6 CONCLUSIONS
Locations that were measured by existing WPS
projects had the accuracy at around Room-level.
However, shadow can provide more precise services
through updates of continuously measured locations
and values from acceleration sensors. In addition, it
remarkably reduces the number of wireless APs
necessary for measurement. The algorithm using the
triangular surveying method requires at least three
wireless APs and more APs to provide higher
accuracy. However, because of the spatial constraint,
corridor, shADow can provide services only through
two wireless APs.
The shADow system reduces the inconvenience
for customers to unavoidably stop to watch
advertisements and cases where customers watch
advertisements that do not induce interest.
With regard to shADow many studies can be
conducted on methods to use probabilities on
algorithms to determine direction changes and noise
compensation.
ACKNOWLEDGEMENTS
"This research was supported by the MKE(The
Ministry of Knowledge Economy), Korea, under the
ITRC(Information Technology Research Center)
support program (NIPA-2012- H0301-12-1006)
supervised by the NIPA(National IT Industry
Promotion Agency).
REFERENCES
C. A. Pomalaza-Raez, C. D. McGillem, “An adaptive,
nonlinear edge-preserving filter,” IEEE Trans. on
Acoustics and Signal Processing, pp. 571-576, 1984.
Edward L. Melanson, Joan R. Knoll, Melanie L. Bell,
William T. Donahoo, J. O. Hill, Lana J. Nysse,
Lorraine Lanningham-Foster, John C. Peters, James A,
Levine, Commercially available pedometers:
considerations for accurate step counting, Preventive
Medicine, vol. 39, pp. 361-368, Aug 2004.
B. Robert, K. Pahlavan, “Site-specific RSS signature
modeling for WiFi localization,” IEEE Global
Telecommunications Conference, pp. 1-6, 2009.
A. Kubber,“ Location based services”, Wiley and Sons,
2005.
Korea Association for ICT Promotion, LBS Technology
and Market Trend report,” 2010.
Hakyong Kim, “A speed-adaptive location estimator for
wireless LAN-based RTLS systems,” The 12
th
Asia-
Pacific Conference on Communications 2006.
P. Bahl and V. N. Padmanabhan, “RADAR: an inbuilding
RF-based user location and tracking system,” in
Proceedings of 19th Annual Joint Conference of the
IEEE Computer and Communications Societies
(INFOCOM ‘00), Vol. 2, pp.775-784, 2000.
Open Mind Common Sense, http://
openmind.media.mit.deu/.
Y. S. Cho, S. Y. Cho, B. D. Kim, S. H. Lee, J. C. Kim, W.
S. Choi, Technical Trend of Indoor/Outdoor Seamless
Positioning,” ETRI, pp. 20-28, 2007.
Playmap, http://www.playmap.co.kr/.
Foursquare, http://www.foursquare.com/.
Paul, A. S., Wan, E. A., “RSSI-Based Indoor Localization
and Tracking Using Sigma-Point Kalman Smoothers,”
Selected Topics in Signal Processing, IEEE Journal of,
pp. 860-873, 2009.
Ekahau, http://www.ekahau.com.
Mianbo Huang, Guoru Zhao, Lei Wang, “A Pervasive
Simplified Method for Human Movement Pattern
Assessing,” 2010 16
th
International Conference on
Parallel and Distibuted Systems, pp. 625-628, 2010.
Consumer Propensity and Location Analysis based Real-time Location Tracing Advertisement Service Design and
Implementation - Real-time Location based Advertisement System
179