Location-sharing Model in Mobile Social Networks with Privacy
Guarantee
Tiago Antonio Rosa and Sergio Donizetti Zorzo
Computer Science Department, Federal University of São Carlos, São Carlos, SP, Brazil
Keywords: Privacy, Mobile Social Networks, Rules of Privacy, Location-sharing.
Abstract: Mobile social networks allow users to access, publish, and share information with friends, family, or groups
of friends by using mobile devices. Location is one kind of information frequently shared. By using
location-sharing on a social network, users allow service providers to register this information and use it to
offer products and services based on the geographic area. Many users consider offers a personal gain, but
for others, it causes concerns with security and privacy. These concerns can eliminate the use of mobile
social networks. This paper presents a model of a mobile social network with a privacy guarantee. The
model enables the user to set rules determining when, where, and with whom (friends or a group of friends)
location information will be shared. Moreover, the model provides levels of privacy with anonymity
techniques which hide the user’s high-accuracy current location before it is shared. To validate the model, a
mobile social network prototype, MSNPrivacy (Mobile Social Network with Privacy), was developed for
Android. Tests were carried out aiming to measure MSNPrivacy’s performance. The results verify that the
rules and privacy levels in place provide an acceptable delay, and the model can be applied in real
applications.
1 INTRODUCTION
Mobile social networks have grown immensely due
to hardware improvements and the reduction of costs
to buy mobile devices. This is verified through
Facebook, which boasts 1.4 billion users worldwide
who are frequently checking the latest updates and
sharing interests with friends and family (Statistic
Brain, 2014). Mobile devices represent a simple way
to share information, such as location, with other
people.
By using location-sharing on a mobile social
network, the user not only provides his current
position to friends and family, but also allows the
net to register this location information. This
information enables the social network to offer and
advertise products and services near his
geographical area.
Location-sharing enables the social network to
offer recommendations, which some users consider
beneficial, while others find it a privacy concern.
Furthermore, malicious users can easily join the
social network and, in this way, get locations of
other users. In possession of this information, these
malicious users can pose a serious threat to the
privacy of other users. Evidence of such concerns
can be seen in related works (Toch et al., 2010);
(Benisch et al., 2010), and they can affect the users
initiative in using these applications. Some mobile
social networks, like Facebook, already offer
privacy control to their users when using a certain
application. This control is carried out with respect
to established privacy policies that determine which
social network users can know a person’s location
and in which places, dates, and times this
information can be published. Such policies
established by social networks, when existing, are
implemented with limitations in relation to offering
the real user location and the group of users.
This paper presents a location-sharing model for
mobile social networks with a location privacy
guarantee for individual users and groups of users.
Users can customize privacy through levels and
rules. The levels define the accuracy of the location
which will be shared, and the rules define where,
when, and with whom the user location will be
shared. The motivation for the development of this
model was to minimize the risks to users’ privacy in
the use of Social Networks Furniture. Such risks
allow users’ identities to be gauged through the
44
Antonio T. and Donizetti Zorzo S..
Location-sharing Model in Mobile Social Networks with Privacy Guarantee.
DOI: 10.5220/0005383200440054
In Proceedings of the 17th International Conference on Enterprise Information Systems (ICEIS-2015), pages 44-54
ISBN: 978-989-758-098-7
Copyright
c
2015 SCITEPRESS (Science and Technology Publications, Lda.)
location; moreover, their safety may be
compromised, as the location history allows to
obtain the path performed.
The proposed model was validated through
implementation of a prototype for a mobile social
network called MSNPrivacy (Mobile Social
Network with Privacy). MSNPrivacy allows the user
to share his location with members of his social
network or with groups created by the user himself
or those created automatically, taking into account
certain contexts such as the degree of relationship or
the geographic position.
Tests were carried out with the aim of measuring
the model’s performance. Results show the delay
provided by privacy techniques is acceptable, and
the model not only has a satisfactory answer time,
but can be implemented in real applications.
The paper is organized as follows: Section 2
presents a discussion about related works; Section 3
presents a privacy context for location-sharing in
mobile social networks; Section 4 presents the
proposed model and its architecture; Section 5
presents the mobile social network prototype
implemented from the presented model,
MSNPrivacy; Section 6 presents the results obtained
in performance tests; Section 7 presents the results
obtained in usability estudy and; Section 8 presents
the conclusion and future works.
2 RELATED WORKS
A great effort has been made by several authors on
the subject of information-sharing privacy in mobile
social networks. Some of these works were
imperative to developing the model proposed in this
paper.
Smith et al., (2005) carried out an initial
investigation on technologies which allow people to
share their location in mobile social networks. The
result of this work was the development of a system
called Reno. This system allows its users to
manually share their location with other people and
to pre-define locations or regions. Reno uses SMS to
notify the location inside the social network, and the
coordinates are obtained through the triangulation of
cell phone towers. These technologies were used due
to the high cost of the most modern devices with
GPS. The privacy controls in the Reno system are
performed only when the user decides to share his
location. The proposed model in this paper, besides
considering when the user wants to share his
location, privacy levels are offered, with each having
a defined characteristic to guarantee the high-
accuracy location remains hidden.
Toch et al., (2010) presented Locaccino, a
location-sharing application for mobile social
networks which allows the user to define rules to
publish his location information. These rules are
defined considering the person, the time, and if the
user is willing to share his current position.
However, the shared location is high-accuracy
information. The proposed model, besides
considering the aspects approached in the Locaccino
application and also allowing for the definition of
rules, applies algorithms which guarantee the current
location’s anonymity.
Bilogrevic et al., (2013) investigated users’
preferences in sharing their location in mobile social
networks. The authors made use of users’ behaviors
when sharing their locations, and the result obtained
was that users cannot specify correctly their
location-sharing preferences. From that information
came the creation of SPISM, a location-sharing
system that (semi) automatically selects which rule
must be applied when a user receives a request to
share his information. Rule selection is made using
machine learning algorithms. However, as with the
other works, SPISM publishes high-accuracy
location information. The proposed model does not
use machine learning algorithms to select the best
rule for the location-sharing requirement. Instead, it
enables the user to manually set the rules to be
applied with each user or one of his groups.
Moreover, the model allows the user to match his
sharing preferences with levels which apply
algorithms to hide the high-accuracy location before
it is shared.
Ribeiro and Zorzo (2009) presented the LPBS
(Location Privacy-based System), a system based on
levels which guarantee users’ location privacy in
LBS. This system is divided into levels, with each
level having distinct privacy characteristics that
guarantee high-accuracy location anonymity.
However, while the LPBS guarantees users’ privacy
in LBS, it is not applied in mobile social networks.
The proposed model uses levels which guarantee a
user’s location anonymity and matches these levels
with publishing preferences.
3 PRIVACY
Privacy concerns were not caused by the emergence
of computers and the internet, rather, they existed
long before. Computers, the internet, and the wide
storage of data it enabled made it possible to collect,
process, and transmit large volumes of data,
Location-sharingModelinMobileSocialNetworkswithPrivacyGuarantee
45
including personal data. Current privacy studies
generally involve different dimensions: law, ethics,
and information technology (Benisch et al., 2011).
The most accepted definition of privacy is the
right of users, groups, or institutions to determine
when, how, and to what extent their information is
communicated to others (Bilogrevic et al., 2013).
This definition captures the idea that privacy is not
simply the lack of information about the user, rather,
it is the control users must have of information
related to them.
Location privacy is defined by Xu et al., (2010)
as a special kind of information privacy concerning
an individual’s claim for determining when, how,
and to what extent location information about them
is communicated to others (Toch et al., 2010). The
person whose location is being measured must have
control of when it is published and who can view it.
Access to user location information must be
available only as per a user’s privacy preferences
(Toch et al., 2010).
In this paper, privacy is understood as the user’s
right to use mobile social networks and define
where, when, and with whom his location will be
shared, as well as the accuracy of this information.
When using mobile social networks, the user
may share his interests and location with friends,
family, and co-workers. However, with the
increasing use of these applications, users’ concerns
related to privacy issues have also increased.
All of the content posted, especially the location,
can be accessed or shared by other entities, which
represents the origins of privacy attacks. Gao et al.,
(2011) considered these origins as gaps; they can be
classified as follows: (i) other users of the social
network, (ii) third-party applications, and (iii) the
actual service provider.
Other users of the social network pose a threat
due to the ease that attackers have in joining the
social network, creating an account on the service
provider and becoming an authentic user. These
malicious users can also produce false context
information for the RSM to gain access to resources,
gain a temporary identity, and to integrate user
groups that are automatically formed by the Mobile
Social Network.
Third-party applications are developed using the
Application Programming Interface (API) available
from the Mobile Social Network. The API provides
an open interface for the creation of new resources
in the Mobile Social Network. These applications
are developed by third parties and therefore are not
always reliable. Third-party applications can be
games, music applications, photos, videos, and
applications used by advertising agencies to carry
out campaigns promoting products and services.
These applications grant the user free access
permission to personal information such as shared
locations, move history, etc.
The social network service provider is
responsible for providing the necessary resources to
users of the Mobile Social Network. This person has
access to all of the personal information relating to
users who have been issued or published. For this
reason, users are left to rely on the service provider
and are becoming more and more concerned about
not really knowing who is manipulating their
information.
Previous works (Benisch et al., 2011); (Smith et
al., 2005); (Bilogrevic et al., 2013) show that users
feel more comfortable when they know who they are
sharing information with. These preferences can be
easily reached through a white list or creation of a
list of people the users would be willing to share
their information with. However, a user’s
preferences are beyond the simple use of these lists
(Benisch et al., 2011).
In 2009, Toch et al., (2010) noted that users have
common preferences such as, “I am willing to share
my location with my friends during the week, nine
to five, and only when I am at my workplace”. This
consideration can be established by rules in the
application that is being used or the human computer
interaction or in an intermediate mechanism device.
Research has been carried out to try to discover
patterns of behaviors and location-sharing
preferences (Benisch et al., 2010); (Toch et al.,
2010). These research results show that privacy goes
from simple to the most complex. The resulting
preferences are listed below (Benisch et al., 2010):
White Lists: Allow users to indicate specific
members of their social network with which they
feel comfortable sharing their location. This is
the simplest privacy definition considered.
Places: Allow users to indicate specific places
where they feel comfortable sharing their
location with members of their social network.
This preference is considered more complex than
white lists.
Times: Allow users to indicate the time period
they feel comfortable sharing their location with
members of their social network. As with place
preference, time preference is more complex
than the white list.
Days: Allows users to indicate which days of the
week they feel comfortable sharing their location
with members of their social network.
ICEIS2015-17thInternationalConferenceonEnterpriseInformationSystems
46
Users can match the preferences presented above,
increasing the complexity further.
The model proposed in this paper encompasses
the preferences described above and enables users to
create privacy rules using one or more preferences.
Moreover, the model provides levels which provide
accuracy and anonymity of the location to be shared.
Matching the preferences and the privacy levels, a
user can reduce his risks when sharing location
information.
4 MODEL
Figure 1 illustrates the mobile social network model
proposed in this work which provides a privacy
guarantee. It is composed of a client application
working in mobile devices, a social network server
(SNS), a reliable proxy, and users.
The application working in mobile devices must
be able to connect to the internet through Wi-Fi
networks or a data plan from operators, connect to
servers, and determine its location through GPS,
triangulation, and a Wi-Fi signal. Moreover, the
devices must be able to receive location
requirements and process rules and techniques
which guarantee users’ privacy.
Figure 1: Model architecture.
The social network server, illustrated in figure 1
by Mac B, meets user requests. The main goal of the
server is to allow users to find the current IP address
of social network members whenever they require
send a request of location. For that, the server stores
a list of its users, their contact lists, and the updated
IP addresses of each user. Moreover, the server
stores the current status of users. Users interact with
the server when registering (only once), during the
login phase (every time a user connects or
disconnects), when downloading a list of contacts,
when periodically informing the updated IP address
and its status, and when sending location
requirements to a member of their net.
In the model, users can be classified in two
different ways: a requester, when sending a request
to another user, or a target, when receiving a
location-sharing request.
To increase communication safety, all requests
made between users and the social network server
are encrypted with a public key certificate obtained
from a reliable Certification Authority (CA).
To protect users’ location privacy in relation to
the server, none of the privacy techniques are
applied by the server, but instead by the application
in the users’ devices. The information is sent to the
server only after necessary alterations; this way, the
server cannot know the real user location. This is a
crucial aspect of the model, as it does not permit the
server, which can be outsourced, to have any
information relating to a user’s location. However,
the server knows their IP address and can, therefore,
accurately infer their location (based on the IP-
geolocalization). To solve this problem, users can
hide their IP address using a reliable proxy which is
illustrated in figure 1 by Mac A.
In this model, the privacy rules defined by the
user consider the position, period (in hours), days of
the week and whitelists. Other rules may also be
incorporated in the model.
The main goal of the model is to guarantee the
privacy of users who share their location with
members individually or with an existing group in
their social network. For that, besides meeting the
users’ preferences presented above, the model
implements mechanisms which guarantee privacy
through anonymity and the hiding of high accuracy
positions. The model offers privacy on three
different levels, allowing the user to personalize
privacy settings according to his needs and interests.
The details of each level are described as follows.
4.1 Level 1 – Accuracy Adjustment
The accuracy adjustment technique (Ribeiro and
Zorzo, 2009) consists of modifying the high
accuracy location so that this information about
position represents different points inside a certain
area. This adjustment is calculated based on the
original location’s random displacement for any
direction inside a given radius, as shown in figure 2.
In Figure 2, the real user location is illustrated by
the center circle A; the displacement radius is
illustrated by the letter ‘r’. When the accuracy is
Location-sharingModelinMobileSocialNetworkswithPrivacyGuarantee
47
adjusted, the user’s location is randomly displaced
for any direction inside the area illustrated by the
smaller circle B. The displacement radius or
accuracy adjustment coefficient is defined by the
user according to his preference.
To perform the accuracy adjustment, the formula
below was used. It is based on the spherical triangle
formed by the initial point, the final point, and the
north pole (Hochbaum and Shmoys, 1985); (Smart,
1977).
Two sides of this triangle are already known: the
side between the initial point and the north pole and
the side referring to the displacement distance. In
addition, the angle formed by these two sides is also
known, and it is characterized by the displacement
direction.
2 arcsinsin
∗cos
cos
1
∗
sin
∗cos)
2sin
∗sin
∗cos
1
,cos
sin
1
∗sin
2
2
1,2∗
In the formula above, the θ element represents the
displacement direction where lat1 and lon1 are the
coordinates of the initial point, and the angular
distance is illustrated by d/R, where d is the
adjustment distance set by the user and R is the
Earth’s radius. The goal of the module applied in the
end of the formula is to accommodate cases in which
the points are in opposite meridians. The resulting
coordinates are illustrated by lat2 and lon2. This
formula is valid only for cases in which the
calculated distance is less than a quarter of the
Earth’s circumference.
Figure 2: Accuracy adjustment.
After performing the accuracy adjustment in the
user location information, the application sends the
resulting coordinate to the SRS.
This technique guarantees that user location
information shared with friends or a group of users
inside a social network will not be a high accuracy
location, and in this way, level 1 guarantees a certain
amount of privacy to users.
Users can still match the accuracy adjustment in
level 1 with the privacy preferences presented in
Session 2 specifically to place, day of the week, and
break time.
4.2 Level 2 – Anonymity
The model in level 2 guarantees user anonymity
through a location information hiding technique. To
hide information, information about the location of
online users who are geographically near is used. In
the case of a group defined inside a social network,
the coordinates of the users in the group are used. To
obtain the necessary locations so the anonymity
technique can be applied, the application requests
the locations of other users.
The hiding location is calculated by selecting one
of the online groups’ locations. This selection is
made by calculating the fair point, a location which
reduces the distance of any user related to all others.
For calculating the fair point, the k-center problem
was used. In the k-center problem, the goal is to find
the k site among all locations shared so that the
maximum distance of any user in relation to the
others is reduced. Figure 3 illustrates an example of
a scenario modeled with the k-center problem,
where the fair point is calculated with four users.
In figure 3, the traced lines represent the
maximum distances, while the full line represents
the minimum distance among all the maximum
distances. Therefore, in this scenario, the fair point is
User 2.
After selecting the location which will hide the
user’s real position, the accuracy adjustment
technique in level 1 is applied in the selected
location and shared after being changed. The
accuracy adjustment is necessary because the
location which will be shared is a high accuracy
location and is a valid location of a social network
user.
The anonymity technique of this level cannot
always be applied because it depends on the
locations of other users within the group. If these
locations are not obtained within five seconds, the
application fine-tunes the current position of the user
and conceals the actual location of the user.
ICEIS2015-17thInternationalConferenceonEnterpriseInformationSystems
48
Figure 3: Scenario modeled with the k-center problem.
4.3 Level 3 – Anonymity Guaranteed
In level 3, the model guarantees the existence of the
anonymity group through the false location
technique (Ribeiro and Zorzo, 2009), as illustrated in
Figure 4.
Figure 4: Generation of false locations.
The client application gets the current position of
the user, shown in Figure 4, at circle “a” using the
device’s GPS. From this information, the application
generates four other false locations, shown in Figure
4 by dark circles, using the characteristics of the
level of precision adjustment technique 1. After this
process, the application has five locations (four false
and the current user). Then, the same technique is
applied to compute the level 2. Anonymity is also
applied at this level. The resulting location of
anonymity, represented by circle “b” in Figure 4, is
fine-tuned and then shared in the social network.
The model guarantees the user’s privacy because the
shared location will not be his real coordinate.
All techniques applied are performed on the
user’s device itself to guarantee the server cannot
obtain the user’s real location; the shared location is
not real.
Users can even combine privacy levels with their
sharing preferences.
The proposed model was validated by the
implementation of a prototype for a mobile social
network called MSNPrivacy (Mobile Social
Network with Privacy). MSNPrivacy allows the user
to share his location with individual members of his
social network or with groups created by him or
automatically, taking into account a given context,
for example, the relationship or geographic position.
Details about MSNPrivacy’s operation will be
discussed in the next session.
5 MSNPrivacy
MSNPrivacy(Mobile Social Network with Privacy)
is a mobile social network application prototype
implemented in Android (2014) which was
developed with the aim of validating the proposed
model. In the prototype implementation, the simple
and intuitive natures of social networks were
considered as far as usability. To compose the user’s
mobile social network, the prototype uses
Facebook’s friends lists. Figure 5 shows the main
application interfaces where users can login and
register, see the list of contacts and the status,
require the location and see the feedback, and set
their preferences and privacy levels.
Figure 5: Main MSNPrivacy application interfaces.
5.1 Operation Principle
MSNPrivacy works as follows: First, the user logs in
or registers on the server, providing the access login
and password. The application also provides the
(
1
)
(2)
(3)
(4)
a
b
Location-sharingModelinMobileSocialNetworkswithPrivacyGuarantee
49
option to register directly through Facebook as
shown in figure 5 (1). If the user is connected to
Facebook, the application directly captures the login
information with no need for it to be typed by the
user. At this point, in addition to sending the data to
the server, the application starts a service which
periodically updates the current IP address and also
starts another service which monitors the reception
of location requests from other users of the social
network. After obtaining the user login data, the
application sends it to the server. The server receives
the data and, if the user is not registered, the server
promotes integration with Facebook, looking for the
user’s list of contacts. After obtaining the contacts,
the server stores them and sends them to the user’s
device which locally stores the contact list and
presents it to the user. The process referring to the
user’s first access is presented in the sequence
diagram in Figure 6.
The user can set his preferences and privacy
levels for each contact of his list and for the groups
he is in (figure 5 (4)). If this setting is not performed,
the application uses level 1 as the pattern for all
contacts and groups.
In a typical scenario, the User requests one of his
online contacts’ location by selecting him from his
list. Next, the application prepares the request and
sends it to the server. The server looks for the
requested user’s current IP, connects to the device,
and sends the request. After receiving the request,
the application verifies the privacy preferences set
for the requested user, applies the privacy techniques
on the obtained location, encapsulates the data, and
sends it back to the server. The server directs the
result to the Uses and shows the location on a map
after receiving the information. If the requested user
is not online at that moment, the requested user’s
application comes back to the main interface, and
when the server receives the requested user’s data, it
sends it to the User’s device. As soon as the
application receives the feedback, it notifies the
User. The requirement information received or sent
is stored either in the server or in the user’s device.
This way, users can audit in the application or on the
web. The process related to the location requirement
is presented in the sequence diagram in figure 7.
All privacy techniques presented in each level
are processed in the user’s device. This ensures the
server does not have access to high-accuracy
location information which is shared on the social
net. However, the server still has the user’s IP
address which makes it is possible to infer his
location. This issue can be easily solved using a
reliable proxy as long as one is available.
Figure 6: First access sequence diagram.
Figure 7: Location request of a contact.
5.2 Privacy Preferences Setting
The application allows the user to set his privacy
preferences with each member of his social network
and within the groups he has joined (figure 5 (4)).
Moreover, the user can match his preferences with
the privacy levels offered.
The proposed model allows to define rules. In
MSNPrivacy some rules such as "Who", which
allows the user to indicate which member of its
mobile social network feels the urge to share his
position, "Time", which allows to determine the
period of schedule was implemented, "Where",
which to determine the location and "days" which
enables want to determine the day or days of the
week that your location will be disclosed.
To better illustrate the privacy setting, a
hypothetical user called “Alice” will be considered.
Alice wants to share her location with friends in her
social network when she is at home and on the
weekends from 9:00 a.m. to 5:00 p.m. In this case,
when creating a setting, Alice will have to indicate
the location, the period of time, and the days of the
week. Alice can even specify the privacy level. In
case no level is specified, level 1 is automatically
used. Alice will be able to create a rule for each
member or group (individually) in her social
network or set a rule for everybody.
ICEIS2015-17thInternationalConferenceonEnterpriseInformationSystems
50
The model deals with the conflict rules in two
ways. In the first case, the model allows the user to
only create rules that grant sharing your location and
does not allow the creation of rules that deny
sharing. For example, “Mary can see my location
from 9 AM to 5 PM,” but she is not allowed to
specify, for example, that “my friends cannot see my
location on the weekends.” In the second case, the
rule is checked before it is committed. This check is
performed to prevent the rules from having
conflicting parameters. For example, “Mary creates
a rule that allows your friends to see your location
on the weekend,” so she failed to create a new rule
that specifies that “your friends can see your
location on weekends from 9 AM to 5 PM.”
6 RESULTS
Aiming to measure the model’s performance, tests
on the main options available in each level were
carried out, such as the first access, where the
integration with Facebook took place, location
request, and the privacy techniques.
To perform these tests, the Android operational
system emulator with IDE Eclipse and the ADT
plugin were used to perform the application
prototype. For the provider, a machine with a
Windows 7 operational system (2GHz CPU, 3GB
RAM) was used.
Figure 8 presents the model’s performance when
the user registers. The result indicates that running
time increases in correspondence with the number of
contacts the user has on Facebook. This happens
because the moment the user registers, the
application makes an authentication with the server.
The server then promotes integration with Facebook
with the goal of looking for and registering all the
user’s contacts that are popular in his mobile social
network. Runtime time can also vary according to
the internet connection used. As this action is carried
out only when the user registers, it does not disturb
the model’s performance.
The following results show the model’s
performance for each privacy level.
Figure 9 presents the model’s performance when
applying the accuracy adjustment in level 1. As can
be seen, the average runtime is constant, taking 0.3
seconds. This is because the accuracy adjustment
calculation is always applied on the user’s individual
location and not in a group of users. Therefore, the
number of users does not have an influence on the
outcome. What can impact running time is
communication with a GPS and the internet
connection quality.
Figure 8: Runtime time during user registration.
Figure 10 shows the model’s performance when
applying privacy techniques in levels 2 and 3. As
can be seen, as the number of users in a group
increases, the runtime also increases. This happens
because to have anonymity, the algorithm in these
levels uses location information from all users in the
group.
Figure 9: Level 1 Runtime.
The difference in the runtime in level 3 occurs
because, in this level, besides calculating the anonymity,
the MSNPrivacy generates false locations using the
accuracy adjustment and applies this same technique in the
user’s high-accuracy location.
As in the other results, the runtime in levels 2
and 3 can change depending on the internet
connection quality or the GPS sign quality.
Moreover, depending on the privacy rule set by the
user, the running time can increase. However, this
time does not put the model performance at risk.
Figure 11 illustrates the runtime results of the
tests with rules. Tests to measure the performance of
the model were realized based on the number and
type of rule. These tests considered the worst-case
scenario: one rule configured with all of the possible
parameters and all of the rules with all of the
possible parameters. In this test, we took three
measurements based on the number of rules (10,
100, and 1,000 rules) configured with all of the
0
2
4
6
8
10
12
14
16
18
20
500 700 900 11001200130014001500
Time[seconds]
No.ofcontactsonFacebook
Registrytime
Location-sharingModelinMobileSocialNetworkswithPrivacyGuarantee
51
possible parameters. The average time spent with 10
rules was 14 seconds; with 100 rules, the average
time was 25 seconds; with 1,000 rules, the average
time spent was 60 seconds. These results were
plotted on a graph in Figure 11 (new version). In this
figure, we draw the trend line to get the behavior of
the system performance based on the number of the
rules. Analyzing these results, we conclude that for a
set of rules (n) to infinity, time behavior tends to
increase linearly (i.e., the model runtime with no set
rules is O (n)).
Figure 10: Runtime in levels 2 and 3.
Figure 11: Runtime time x number of rules.
Based on the results obtained from the tests
carried out in the privacy levels and testing with the
rules, the average delay added by the implemented
prototype was 35 seconds. This delay time is
proportional to the implementation of level 2 and 3,
with 10 users per group. The inserted delay can vary
according to the number of users in a group. In a
group with more than 100 users, the implementation
of level 2 can become unwieldy. The solution to this
problem is to adjust the amount of locations used in
the calculation of anonymity. Level 3 is not
compromised in this case because the locations used
are generated by the application; the initial number
of points generated is four.
Although a delay is inserted in the model
performance, it is safe to say this delay does not
damage the usability of the mobile social network,
since the delay will only happen when the user
receives a request for publishing his location. With
the other functionalities, there will be no delays in
performance. Furthermore, test results show that the
delays generated by the privacy techniques offered
are at proportional and justifiable levels; further, the
privacy rules are desirable and chosen by users.
7 USER STUDY
A usability study was conducted to get the users’
perceptions of and measure the effectiveness of the
privacy mechanisms the model offered. The study
was conducted with 50 participants, aged 19 to 30 in
the exact sciences, health, and humanities. A
previous study found that many of the problems that
are likely to occur in a given population were only
identified by five participants (Leon, 2012). We
asked the participants to answer questionnaires. The
issues addressed in the questionnaires were designed
to measure user satisfaction related to privacy and
usability.
The study consisted of two phases. In the first
phase, participants answered an initial questionnaire
composed of nine questions. The objective of the
first questionnaire was to measure the level of
participants’ understanding of privacy in mobile
social networking. In addition, the responses from
this survey will be compared with the responses to
the second questionnaire. In the second phase,
participants used the main features of the prototype
to create rules and privacy levels, location, and to
request access to audit. Then, we asked participants
to answer the second questionnaire, which contained
21 questions. The objective of this phase was to
obtain feedback from participants on the usability
and efficiency of the privacy techniques offered. A
five-point Likert scale (where one means strongly
agree and five means strongly disagree) was used.
The results of the study show that, at the
beginning of the tests, 77% of users had mistaken
understanding of privacy in mobile social
networking or did not worry about privacy issues.
The results at the end of the tests show that 100% of
the participants believed that privacy safeguards
were important and influenced location-sharing
decisions in mobile social networking. Regarding
the privacy techniques the model offered, 87% of
participants approved of its efficiency, while the rest
of the participants (13%) did not approve or were
indifferent. On the usability of the prototype, 76%
found it difficult to configure privacy rules and
levels the way the prototype offered. Overall, 80%
of participants would use the mobile social network
prototype in their day-to-day lives.
ICEIS2015-17thInternationalConferenceonEnterpriseInformationSystems
52
8 CONCLUSION AND FUTURE
WORKS
Mobile devices represent a simple way to quickly
share information with other people and the use of
mobile social networks is a easy way to share
location.
When a user shares his location with friends and
family on a social network, he also allows this
information to be registered by the social network
itself, which then uses it to offer products and
services according to location. For many users, this
offer is seen as a benefit, but for many, privacy
concerns outweigh that benefit. These concerns can
reduce motivation to use social networks.
The privacy offered by the model through the
definition of rules and levels prevents users from
attacks mentioned in section 2. As the shared
location information are hidden throughout the
levels, malicious users do not get the information
real position and with high accuracy user. Thus,
hinders your identity is inferred by location. In
addition, the social network provider does not store
the history of shared locations and thus cannot to
trace the stream of the users. The other members of
the social network store all locations shared your
friends, however, is very difficult to get the user path
if it has set policies and levels for the share.
An interesting point to note is that the model
does not fully committed services and
recommendations offered by the social network,
therefore users allow their location to be shared
according to the rules. The location modified by
levels still belongs to the geographical area where
the user is. Thus, services and offers
recommendations based on geographic area can still
be made. But personalization is the user's
responsibility, since it allows its position to be
shared, services and recommendations made by the
social network will not be compromised.
The results show that the model has good
performance, despite the existence delays. Analysing
the worst case, the model runtime to process an
incoming request is on average 28 seconds. This
delay added to the implementation of the social
network is acceptable when compared to the average
time spent by the GPS to obtain the first location
that is 30 seconds. The test results also show that
delays generated by using the privacy techniques are
proportionate to the desired levels of privacy of user.
The results obtained in the evaluation of users
show three important points: First point is that users
believe that, for the use of mobile social networks,
supply and privacy safeguards is essential. The
second point is that most users approved the
efficiency of privacy techniques offered by the
model. However, despite the privacy of efficiency is
approved, the prototype needs improvements over its
usability.
Among the future works are: The need to move
employees algorithms in the privacy level 2 in order
to improve application performance; The way users
perform the privacy rules setting should be
optimized using the results of the usability study;
Evolve the prototype is makes it the closest to an
application of real mobile social network and,
finally, conduct a study to measure the configuration
preferences user privacy.
REFERENCES
http://www.statisticbrain.com/social-networking-statis
tics/; last access: August, 2014.
Toch, E., Cranshaw, J., Hankes-Drielsma, P., Springfield,
J.; Gage, P., Cranor, L.; Hong, J., Sadeh, N., 2010.
Locaccino: A Privacy-centric Location-sharing
Application. In 12th ACM International Conference
Adjunct Papers on Ubiquitous Computing, pp. 381-
382.
Benisch, M.; Kelley, P. G., Sadeh, N., Cranor, L. F., 2011.
Capturing Location-Privacy Preferences: Quantifying
Accuracy and User-Burden Tradeoffs, In Personal and
Ubiquitous Computing; vol. 15, pp. 679-694, 2011.
Toch, E., Cranshaw, J., Drielsma, P. H., Tsai, J. Y.,
Kelley, P. G., Springfield, J., Cranor, L., Hong, J.;
Sadeh, N., 2010. Empirical Models of Privacy in
Location-sharing. In 12th ACM International
Conference on Ubiquitous Computing, pp. 129-138.
Smith, I., Consolvo, S., Lamarca, A., Hightower, J., Scott,
J., Sohn, T., Hughes, J., Iachello, G., Abowd, G. D.,
2005. Social Disclosure of Place: From Location
Technology to Communication Practices. In Third
International Conference on Pervasive Computing, pp.
134-151.
Ribeiro, F. N., Zorzo, S. D., 2009. LPBS – Location
Privacy Based System. In IEEE Symposium on
Computers and Communications, pp. 374-379.
Shin, K. G., Ju, X., Chen, Z., Hu, X., 2012. Privacy
Protection for Users of Location-Based Services. In
IEEE Wireless Communications, vol. 19, pp. 30-39.
Harrison B., Dey, A., 2009. What Have You Done With
Location-Based Services Lately?. In IEEE Pervasive
Computing, vol. 8, pp. 66-70.
Bilogrevic, I., Huguenin, K., Agir, B., Jadliwala, M.,
Hubaux, J. P, 2013. Adaptive Information-sharing for
Privacy-aware Mobile Social Networks. In 2013 ACM
International Joint Conference on Pervasive and
Ubiquitous Computing, pp. 657-666.
Xu, F., He, J., Wright, M., Xu, J., 2010. Privacy Protection
in Location-Sharing Services. In International
Conference on Computer Application and System
Location-sharingModelinMobileSocialNetworkswithPrivacyGuarantee
53
Modeling, ICCASM.
Android; http://www.android.com; Last access:
November, 2013.
Dierks, T., Rescorla, E., 2013. The Transport Layer
Security (TLS) Protocol. RFC (Request for
Comments) 4346. Available: http://www.ietf.org/
rfc/rfc4346.txt. Last access: November, 2013.
Leon, P. et al. Why Johnny can't opt out: a usability
evaluation of tools to limit online behavioral
advertising. In: SIGCHI Conference on Human
Factors in Computing Systems. New York, NY, USA.
Pages 589-598. 2012
Hochbaum, D. S., Shmoys, D. B., 1985. A Best Possible
Heuristic for the K-center Problem. Math. of Oper.
res., Vol. 10, No. 2, pp. 180-184.
Smart, S. W., 1977. TextBook on Spherical Astronomy.,
6. ed., England: Cambridge University Press, 415 p.
Williams. Ed. Aviation Formulary 1.44. Available:
http://williams.best.vwh.net/avform.htm#LL; Last
access: November, 2013.
Gao, H. et al. Security issues in online social networks. In:
IEEE Internet Computing, 15(4), pp. 56–63. 2011
ICEIS2015-17thInternationalConferenceonEnterpriseInformationSystems
54
