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

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

 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

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 person’s 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)

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%

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

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.

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

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