Assessing the Cooperativeness of Users in Wi-Fi Networks
Szymon Szott, Grzegorz Ptaszek and Lucjan Janowski
AGH University of Science and Technology, 30-059 Krakow, Poland
Keywords:
Wi-Fi, Network, Selfish, Cheating, Misbehaviour.
Abstract:
Wi-Fi networks are based on the cooperation of users in sharing a common resource – the radio channel. This
is a security risk because users may behave selfishly to increase their own throughput but at the same time
decrease the overall network performance. Many scientific analyses have focused on this problem, but none
have taken into account real user behavior. We present the initial results of a work-in-progress in which we
studied a group of users in terms of their online behavior as well as their psychological characteristics. We
have found that users behave selfishly in a wireless setting, regardless of their cooperative nature. We provide
lessons learned as well as pose open questions for further research in this field.
1 INTRODUCTION
Wi-Fi networks are a popular means of wireless com-
munication: they can be found in homes, offices, and
public places. These networks are based on the prin-
ciple of users sharing a common resource the ra-
dio channel. This need for cooperation leads to cer-
tain security issues, which are known in the literature
as selfish attacks (the terms cheating and misbehav-
ior are also common). These attacks are specific be-
cause they are insider attacks, i.e., they are performed
by users which have already gained access to the net-
work. Selfish attacks are becoming a problembecause
the standard which defines the communication proto-
col for Wi-Fi networks (IEEE 802.11) contains no in-
centives for users to cooperate (Szott, 2014). In fact,
manufacturers exploit this trait to increase the perfor-
mance of their devices (Bianchi et al., 2007).
A prominent example of such behavior is a traf-
fic remapping attack (Konorski and Szott, 2014), in
which a user takes advantage of the quality of ser-
vice (QoS) traffic prioritization mechanism of 802.11
(Natkaniec et al., 2013) and assigns high priority
identifiers to regular, best effort traffic (Figure 1).
This means that regular traffic (e.g., a file transfer) is
treated as if it required low delay (as, e.g., a Skype
call), thus disturbing the operation of the network.
This attack is relatively easy to perform as it requires
adding only one rule in the user’s firewall software.
Non-cooperative behavior has been well-studied
in wireless communications literature, both in terms
of the potential benefit to the misbehaving user (Szott
et al., 2010) as well as countermeasure methods (Szott
et al., 2013a; Szott et al., 2013b). However, de-
spite the multitude of theoretical analyses, practical
user behavior has not been studied in real world Wi-
Fi deployments. This raises the question: Are users
willing to cooperate in a wireless setting? Com-
parable studies conducted for peer-to-peer networks
show that this is not the case (Anagnostakis et al.,
2006). Therefore, we propose the following hypoth-
esis: users of Wi-Fi networks will, given the chance,
exhibit non-cooperative behavior regardless of their
personal character. Towards this end we conducted a
study in which we compared the online behavior of
users (in a simulated environment) with the outcome
of several psychological tests. The initial results are
promising and we report several lessons learned. To
the best of our knowledge, this work in progress is the
first reported study of this kind for Wi-Fi networks.
2 METHODOLOGY
Our study was conducted separately for each partic-
ipant. It consisted of two parts. First, we assessed
their online behavior in a simulated test. Then, we
determined their overall willingness to cooperate us-
ing psychological surveys.
2.1 Online Behavior Test
The participants were provided with a laptop and
asked to test a new application for transferring data
60
Szott S., Ptaszek G. and Janowski L..
Assessing the Cooperativeness of Users in Wi-Fi Networks.
DOI: 10.5220/0005117300600064
In Proceedings of the 11th International Conference on Wireless Information Networks and Systems (WINSYS-2014), pages 60-64
ISBN: 978-989-758-047-5
Copyright
c
2014 SCITEPRESS (Science and Technology Publications, Lda.)
Wi-Fi Network
Voice
queue
Best Effort
queue
MAC frame with
VO AC code
IP packet with
BE DSCP code
User Application
Packet mangling software
EDCA queues and scheduling
Wi-Fi Station
Figure 1: Example of a traffic remapping attack performed by a user of a Wi-Fi station taking advantage of traffic differentia-
tion provided by the enhanced distributed channel access (EDCA) function of IEEE 802.11.
traffic (Figure 2). In reality, the application was only
simulated. The participants were informed of the fol-
lowing: they will be using the university’s Wi-Fi net-
work, there are currently other users in the building
using this network; and the network capacity is lim-
ited and shared among all its users. The user’s task
was to transfer a 50 MB file to a popular social net-
working site. The application automatically config-
ured all necessary parameters. Since the file transfer
took about 5 minutes, users were allowed to browse
the Internet in the meantime. During the transfer, at
fixed intervals, the testing application created a popup
window informing the users that they can increase
their transfer rate at the cost of the rate of the other
users in the network. They then answered a yes/no
question: Do you want to increase your throughput?
This question appeared 10 times throughout the file
transfer. Based on a user’s decision, the application
would modify the transfer rate accordingly. The test
was repeated in two consecutive trials for the up-
link and downlink directions, respectively. The av-
erage throughput values were taken from simulation
studies conducted previously for the uplink (Konorski
and Szott, 2014) and downlink (Szott et al., 2009) di-
rections. In both cases the selfish attack was a traf-
fic remapping attack. This attack increased the traffic
rate approximately two or threefold for the downlink
and uplink directions, respectively.
2.2 Psychological surveys
We applied three different psychological surveys
taken from the literature to assess the level of coop-
eration exhibited by the respondents: the agency and
communion scales (Wojciszke and Szlendak, 2010),
the belief in life as a zero-sum game scale (Rozycka
and Wojciszke, 2000), and the ethic’s questionnaire
(a)
(b)
(c)
Figure 2: Screenshots of the application used for the online
behavior test: (a) introductory screen, (b) application win-
dow during normal operation, (c) notice about the possibil-
ity of misbehavior. Users were asked to read all instructions
carefully.
(Wojciszke and Baryla, 2000). All three tests, pre-
pared in the language of the responders (Polish), have
AssessingtheCooperativenessofUsersinWi-FiNetworks
61
shown to have satisfactory psychometric parameters.
The first scale measures agency (focus on self and
own goals) and communion (focus on other people
and interpersonal relations) as well as unrestrained
agency (excessive focus on self with an ignorance
of social relations) and unrestrained communion (ex-
cessive focus on others with an ignorance of own
agency).
The second scale measures the general belief that
life is a zero-sum game a hidden assumption that
one persons profit or success is only possible at the
cost of an-other person’s loss or failure. People, who
believe that life is a zero-sum game delegitimize the
social system and believe in injustice in the social
world.
The final test measures the degree of faith of the
respondent in two ethical codes: the ethics of auton-
omy and the ethics of collectivism. People with high
scores related to the former are distinguished by re-
spect for the welfare, freedom, and rights of an in-
dividual, helping others, and loyalty to individuals.
People with high scores related to the latter are distin-
guished by respect for the welfare, interest, and rights
of their own group, maintaining group integrity, group
loyalty and conformity.
3 RESULTS AND ANALYSIS
The study was conducted on a group of 20 students
(13 female) from AGH University. The participants
were 21 to 25 years old and familiar with modern
technology. This group of users, while small, nev-
ertheless allowed us to draw initial conclusions and
prepare subsequent research steps in this work in
progress.
The main results of the psychological surveys are
presented in Figure 3. The majority of participants ex-
hibited moderate cooperativeness according to all of
the scales used for measurement. In fact, 70% of the
participants deviated no more than 30% from levels
considered moderate. At the same time the majority
of them refrained from cooperation in a wireless set-
ting: approximately in 80% of cases users answered
yes to the questions posed in the simulated online
behavior test. Furthermore, we analyzed these two
categories of results (psychological survey and sim-
ulated test) using Spearman’s rank correlation coeffi-
cient and found no significant correlation. These fac-
tors confirm our initial hypothesis as stated in the in-
troduction. Further detailed results are described next.
The initial choice of a user describes the first re-
action to the dilemma of cooperation in each trial.
The obtained results confirm the dominance of non-
(a) (b)
(c) (d)
Figure 3: Results of psychological surveys. Participants
level of (a) agency and communion, (b) unrestrained agency
and unrestrained communion, (c) belief that life is a zero-
sum game, (d) faith in ethics of autonomy and ethics of col-
lectivism.
Yes
87%
No
13%
Figure 4: Initial choice of users in each trial.
cooperative behavior (Figure 4). This means that the
majority of participants were willing to cheat from the
very beginning of the test.
We also analyzed the behavior variation of users,
i.e., how often within a given trial each user switched
from one behavior to another (Figure 5). In almost
half of the trials, users did not change their behav-
ior and continued to cheat. Almost as often, users
changed their behavior once or twice. This indicates
that they were willing to experiment with the achieved
throughput and determine what would happen had
they cooperated. Unsatisfied with the lower through-
put, users reverted to cheating. Finally, in the remain-
ing 20% of trials users changed their behavior three or
more times. This indicates a possible further investi-
gation of choice of behavior on the achieved through-
put.
Since each user performed two consecutive trials
(uplink and downlink), we measured how often cheat-
ing occurred in both of them (Figure 6). No statistical
correlation was found, which indicates that the order
WINSYS2014-InternationalConferenceonWirelessInformationNetworksandSystems
62
0%
10%
20%
30%
40%
50%
0 1 or 2 3 or 4 5 or more
Precentage of Trials
Instances
Figure 5: Occurrences of users changing behavior in a trial.
0%
20%
40%
60%
80%
100%
120%
0% 20% 40% 60% 80% 100% 120%
Cheating percentage in downlink
Cheating percentage in uplink
Instances
Figure 6: User behavior in uplink and downlink trials. Bub-
ble size denotes number of instances.
of trials was not important. There were some cases,
where users learned to cheat in the first trial and con-
tinued to do so in the second, whereas other users at-
tempted further experimentation of how their choices
made an impact on the throughput values.
4 LESSONS LEARNED
The lessons learned and open questions brought forth
through this study include:
People, even those with moderate or high lev-
els of cooperativeness, tend to not cooperate in
a Wi-Fi network. This supports our working hy-
pothesis: users of Wi-Fi networks will, given the
chance, exhibit non-cooperative behavior regard-
less of their personal character. Because the sam-
ple population is too low (20 people) to accept
the hypothesis unconditionally, we are motivated
to study whether the results apply in the general
case.
Similar studies should be performed separately
for uncooperative and highly cooperative groups.
This would provide further insight on user behav-
ior. However, due to their intrinsic nature, unco-
operative groups may be difficult to organize.
It would be interesting to study a multi-user envi-
ronment (representative of a hot-spot) to observe
the interactions among participants and see how
the multi-player prisoner’s dilemma plays out.
Different incentive mechanisms may be studied,
e.g., varying the reward for completing the as-
signed task based on time efficiency. Monetary
benefits could be worth considering.
The framing of the question which the partici-
pants are asked in the online behavior test impacts
the results. We opted for a neutral approach (Do
you want to increase your throughput?) because
the use of words such as cheating or misbehaving
might influence the participants’ perception of the
problem, which is also an area of further study.
Having established that non-cooperative behav-
ior in wireless networks is a problem, the design
of countermeasure methods becomes important.
What kind of punishment mechanisms could be
used and would they be effective in incentivizing
cooperation? Can this be done within the frame-
work of 802.11?
This and subsequent studies may also be useful
for service providers, in order to establish their
network strategies (which service parameters can
be decreased without affecting user experience,
how much effort are network users willing to ex-
pend to increase their transfer rate) as well as
for network managers who want to provide QoS-
based resource sharing (Kosek-Szott et al., 2013).
5 SUMMARY
The study of the cooperativeness of Wi-Fi network
users, presented in this paper, has provided initial re-
sults which satisfy our hypothesis: users of Wi-Fi net-
works will, given the chance, exhibit non-cooperative
behavior regardless of their personal character. This
chance is not a purely theoretical concept because of
the emergence of flexible Wi-Fi platforms such as the
Wireless MAC Processor (Szott et al., 2013a). It can
be concluded that, on one hand, the scientific research
performed in the field of wireless network security
and performanceanalysis is justified, and on the other,
that further effort in this work-in-progress is encour-
aged in order to better understand human online be-
havior. Our future research agenda will be based on
the conclusions presented in Section 4.
ACKNOWLEDGEMENTS
This work was supported by the AGH Univer-
sity of Science and Technology under contracts no.
AssessingtheCooperativenessofUsersinWi-FiNetworks
63
15.11.230.051and 11.11.230.018. The authors would
like to thank Lukasz Wronski for his support during
the tests.
REFERENCES
Anagnostakis, K. G., Charmatzis, F., Ioannidis, S., and
Zghaibeh, M. (2006). On the impact of p2p incentive
mechanisms on user behavior. In IN NETECON+IBC.
Bianchi, G., Di Stefano, A., Giaconia, C., Scalia, L., Ter-
razzino, G., and Tinnirello, I. (2007). Experimen-
tal assessment of the backoff behavior of commercial
IEEE 802.11b network cards. In Proc. of INFOCOM.
Konorski, J. and Szott, S. (2014). Discouraging traffic
remapping attacks in local ad hoc networks. Wire-
less Communications, IEEE Transactions on, DOI
10.1109/TWC.2014.2321577.
Kosek-Szott, K., Natkaniec, M., Szott, S., Krasilov, A.,
Lyakhov, A., Safonov, A., and Tinnirello, I. (2013).
What’s new for QoS in IEEE 802.11? IEEE Network,
27(6):95–104.
Natkaniec, M., Kosek-Szott, K., Szott, S., and Bianchi, G.
(2013). A Survey of Medium Access Mechanisms
for Providing QoS in Ad-Hoc Networks. Communi-
cations Surveys Tutorials, IEEE, 15(2):592–620.
Rozycka, J. and Wojciszke, B. (2000). Skala wiary w gre
o sumie zerowej [scale of belief in life as a zero-sum
game]. Studia Psychologiczne, 48:35–46.
Szott, S. (2014). Selfish Insider Attacks in IEEE 802.11s
Wireless Mesh Networks. IEEE Communications
Magazine, 52(6):227–233.
Szott, S., Gozdecki, J., Kosek-Szott, K., Loziak, K.,
Natkaniec, M., and Tinnirello, I. (2013a). The risks
of WiFi flexibility: Enabling and detecting cheating.
In Proc. of Future Network and Mobile Summit.
Szott, S., Natkaniec, M., and Banchs, A. (2009). Impact of
Misbehaviour on QoS in Wireless Mesh Networks. In
Proc. of IFIP Networking.
Szott, S., Natkaniec, M., and Pach, A. R. (2010). An IEEE
802.11 EDCA model with support for analysing net-
works with misbehaving nodes. EURASIP Journal on
Wireless Communications and Networking, 2010:71.
Szott, S., Natkaniec, M., and Pach, A. R. (2013b). Improv-
ing QoS and security in wireless ad hoc networks by
mitigating the impact of selfish behaviors: a game-
theoretic approach. Security and Communication Net-
works, 6:509–522.
Wojciszke, B. and Baryla, W. (2000). Potoczne rozumie-
nie moralnosci: piec kodow etycznych i narzedzie ich
pomiaru [Lay understanding of morality: Five ethical
codes and their measurement]. Przeglad Psycholog-
iczny, 43:395–421.
Wojciszke, B. and Szlendak, M. (2010). Skale do po-
miaru orientacji sprawczej i wspolnotowej [Scales
Measuring Agency and Communion]. Psychologia
Spoleczna, 5:57–70.
WINSYS2014-InternationalConferenceonWirelessInformationNetworksandSystems
64