“MY SMARTPHONE IS A SAFE!”

The User’s Point of View Regarding Novel Authentication Methods and Gradual

Security Levels on Smartphones

Tim Dörflinger, Anna Voth, Juliane Krämer and Ronald Fromm

Deutsche Telekom Laboratories, An-Institute Berlin University of Technology

Ernst-Reuter-Platz 7, 10587, Berlin, Germany

Keywords: Authentication methods, Gradual security levels, Smartphone security, Qualitative user research, Focus

groups, Innovation development.

Abstract: This paper addresses laboratory tests regarding a graded security system on smartphones based on novel

authentication methods. The main scope of this paper is the user’s perception of and the need for such a

system, rather than the technical dimensions of it. In November 2009 we conducted four focus groups with a

total of n=19 respondents with the goal to evaluate different prototypical authentication methods for

smartphones and to determine the effects such methods would have for the user’s interaction with the

devices. The focus groups were part of a larger research program at Deutsche Telekom Laboratories that

included a web survey measuring general user preferences regarding security and smartphone usage as well

as the development of prototypical authentication methods based on Google’s “Android” operating system.

The goal of this research was to integrate the user into the development process as soon as possible and to

determine the overall acceptance of new authentication methods, such as biometric authentication, but also

2D and 3D gestures, recognition based authentication and password authentication. This paper gives

valuable insights on the weakest link of the security chain: the user.

1 INTRODUCTION & FOCUS

In recent years, smartphones

have taken up an

increasingly important part in the lives of

Information and Communication Technology (ICT)

users. Formerly only a gadget to business users and

aficionados or “geeks”, smartphones have set out to

conquer the world, as current sales figures show.

According to estimates from Canalys (2009), the

global market for smartphones was divided into the

five major device manufacturers: Nokia (39.7%),

RIM (20.6%), Apple (17.8%), HTC (5.3%) and

Fujitsu (3.4%) in the third quarter (Q3) of 2009. The

remaining 13.2% were distributed amongst other

At the time this research paper was written, no industry-

standardized definition for smartphone existed. In general,

smartphones are considered to be mobile phones with

extended PC functionality, such as Internet access, data

processing, multimedia capabilities that run on a specific

operating system which allows the development of additional

applications and services.

manufacturers. Compared to Q3 figures of the

previous year, this represents quarterly growth rates

in world wide smartphone sales of over 4% and the

market is far from being saturated. Estimates from

Gartner (2009) show that by 2010 smartphone sales

will make up around 37% of global handset sales. In

addition, their revenue is forecast to reach $191

million by 2012, higher than spendings on mobile

PCs, which is forecast to reach $152 million in 2012.

Today, a constantly increasing number of

consumers use smartphones for a broad variety of

tasks and purposes, ranging from telephony to

instant messaging, from mobile banking to Global

Positioning Satellite (GPS) based navigation.

Smartphones serve as tools for organizing the users’

daily lives through productivity applications such as

calendars, notepads or calculators. Furthermore they

are turned into multimedia toys with capabilities to

play music, videos, and games or surf the World

Wide Web (WWW) and take pictures using

integrated cameras. Smartphones are more than just

communication devices; they are mobile

companions for various situations, ranging from

155

Dörﬂinger T., Voth A., Krämer J. and Fromm R. (2010).

“MY SMARTPHONE IS A SAFE!” - The User’s Point of View Regarding Novel Authentication Methods and Gradual Security Levels on Smartphones .

In Proceedings of the International Conference on Security and Cryptography, pages 155-164

DOI: 10.5220/0002915701550164

 SciTePress

work organizers to holiday planners. But this is only

the beginning. As Dilinchian (2009) acknowledges,

smartphones will also “serve health, emergency

services, defence, education, banking, retailing, and

other sectors benefiting from information services”

in the near future

Smartphone users also store a vast array of

different data on their devices, ranging from

personal pictures to messages (email, SMS, MMS),

contact lists, addresses, birthdates, music, movies

and various other files, depending on the respective

Operating Systems (OS) used. Smartphones can

therefore be considered as light versions of

computers with ubiquitous telephony functionality.

However, this trend also leads to new challenges

in the field of research and development (R&D):

Like desktop computers or their mobile counterparts

such as notebooks or netbooks, smartphones are

increasingly subject to security threats. So far,

relatively little is known about viruses for

smartphones but those and other types of security

breaches will be of relevance in the near future. Due

to the manifold technical possibilities of establishing

connections to the device (UMTS, Infrared,

Bluetooth, W-Lan), the data stored on these devices

is increasingly at risk. In addition, smartphones can

be stolen, lost or merely left at the table unnoticed in

a café or restaurant. If found by a third party with

malign intentions, the data on the device are usually

accessible, unless the owner decided to use

additional device protection mechanisms like pass

codes or device locks.

According to a recent mobile security report

published by McAffee (2009), the “need for

additional security measures on the application,

device platform and network level” arises because

“mobile devices become increasingly multi-

functional and are connected to other guarded and

unguarded networks”. This is primarily because the

main security mechanism with regards to

smartphone or even mobile phone usage still is the

Personal Identification Number (PIN). The PIN

however does not provide sufficient protection to the

data stored on the devices because it authenticates

the user with the network of his Mobile Network

Operator (MNO) when turning on the device. After

the PIN has been entered, the device remains open

and if left in standby mode or constantly turned on,

For some sectors, this is already the case. As the broad

availability of mobile banking and brokerage application

available for the iPhone or Android based phones show, the

finance industry has already benefitted from this development.

no other security barrier prevents third parties from

accessing the device. Unless the user explicitly

chooses otherwise, the default setting for

smartphones is considered to be “insecure”, the

device itself remains like a bike lock, it is either

closed or open.

An increased awareness of these issues has led

some research institutes to the evaluation and

analysis of security concerns from a user

perspective. While technical aspects of data

protection and privacy are without a doubt the final

means to an end, it is of crucial importance for

researchers and developers alike to get an idea of

what users consider necessary with regards to their

communication devices. Only then can technically

adequate solutions be provided, that actually have a

chance of being established on the market. Because

there is still a wide gap between usability

and

security, the combination of both concepts signifies

an important challenge to researchers and developers

alike. If passwords are becoming more complex

users might disapprove because in addition to the

already existing amount of PINs and pass codes that

have to be remembered regularly, e.g. ATMs, email

passwords, PC passwords, etc, the objective increase

in security may easily lead to a subjective decrease

in usability. The longer the password, the more

overburdened the user is and thus experiences a poor

usability. In many cases this can even lead to by-

pass strategies on behalf of the user, such as

choosing the same password or PIN for different

applications or services or opting for passwords that

are easy to remember, such as birthdates or names.

The security of devices is then threatened, because

the user turns out to be the weakest link in the

security chain.

In order to make life easier for the users and

increase the joys of using communication devices, a

solution might be to implement different security

levels on smartphones for accessing different types

of data or applications, thereby increasing the

overall protection. If then combined with novel

authentication methods, such as biometric

authentication or memory-based authentication, a

gradual approach to security, based on individual

preferences, might make smartphone usage safer

Usability, which can be described as the extent to which a

product can be used with “effectivity, efficiency and

satisfaction” (Iso, 1998), plays an important role in the choice

of an authentication method.

For example through a combination of numbers, letters (both

small case and upper case) and symbols.

SECRYPT 2010 - International Conference on Security and Cryptography

156

meanwhile maintaining a manageable level of

complexity.

While more sensitive data are assumed to need

more secure authentication methods, there is another

dimension which has to be regarded: the simplicity

and acceptance of the respective method. It is a

crucial factor regarding the actual usage of these

methods as smartphone owners rarely use methods

which are too complex or which make them feel

ashamed in public. Therefore the usage frequency as

well as the usage context (e.g. being in the subway

vs. being at home or in the office) of applications

has to be regarded additionally when thinking about

a fitting authentication method: A very complex and

intrusive method may be adequate for a very

sensitive and rarely used application, but users will

not choose it for an even more sensitive application

which they have to use very often. Thus, we believe

that the OS of a smartphone should offer various

authentication methods and different security levels,

enabling the user to individually find an optimal

trade-off solution for securing each application.

Since the sensitivity of data or applications is a

subjective assessment which varies between

smartphone users, we decided to address different

types of users in a qualitative research project.

In order to obtain insights regarding the

acceptance of gradual security as well as novel

authentication methods such as biometrics

memory-based authentication methods, we

conducted four different focus groups in Berlin in

November 2009 at our research laboratory. These

qualitative findings were further validated through a

web-survey with 308 smartphone users from two

countries, namely Israel and Germany. The

following Section 2 will describe the methodology

used in the focus groups in order to evaluate novel

authentication methods as well as to test the

acceptance of gradual security. Section 3 will

present the main findings from the focus groups and

partially compare them to the quantitative findings

from the web survey. The final Section 4 will then

summarize and conclude by addressing the

shortcomings of our approach and by highlighting

directions for future research in the field.

Reinhardt, Furnell and Clarke (2009) argue that, although

biometric methods are well-known forms of authentication,

such measures are currently not available on smartphones and

are therefore considered “novel” from the user’s point of

view.

2 RESEARCH METHODOLOGY

2.1 Focus Groups

Focus groups are a qualitative methodological

approach which is often used in product related

consumer research. They can be described as

moderated group discussions which are “designed to

obtain perceptions on a defined area of interest

(Krueger and Casey 2000: 5)”, in our case the

acceptance of gradual security levels and

authentication methods. Focus groups are usually

combined with other methods of empirical social

research. They serve as an interpretative aid and are

therefore not statistically representative. Such

discussion groups normally consist of four to ten

participants, in order to keep the moderation and the

discussion on a manageable level.

The main benefit of a focus group discussion is

to gain qualitative in-depth insights from different

users and their perspectives at the same time.

Participants have the opportunity to discuss freely

and state their point as well as develop assumptions

inspired by the experiences made by others. As

noted by Krueger and Casey (2000), this means it is

not only possible to gain individual opinions about

certain products, services, etc. but also group

specific tendencies. “It is important to recognize that

the amount of direction provided by the interviewer

does influence the types and quality of the data

obtained from the group” (Stewart et al 2007: 38). In

order to retrieve the desired results from such a

group, moderators and observers have to be well

trained to prevent the discussion from shifting in a

wrong direction. The analysis of focus groups is

complicated and time consuming. The usual outputs

of focus groups are video documentations, voice

recordings, drawings made by the participants,

questionnaires and pictures. Besides these collected

data, focus groups are usually conducted in a

laboratory or studio with a semi-transparent “mirror-

wall”, allowing others behind the mirror to observe

the group unnoticed.

Through conducting the focus groups we

intended to unveil qualitative insights regarding

current security requirements and possible

acceptance of alternative authentication methods.

Because of budgeting constraints we recruited a

total of n=19 participants, split across 4 groups.

Table 1 gives an overview of the sample

distribution:

"MY SMARTPHONE IS A SAFE!" - The User's Point of View Regarding Novel Authentication Methods and Gradual

Security Levels on Smartphones

157

Table 1: Overview of the focus group distribution with

regards to socio-demographic criteria.

The participants were clustered into four groups

according to an internal segmentation procedure,

such as parents, students, business professionals and

fully employed singles and couples. After giving an

overview of the types of users we recruited for our

tests, we will now discuss the topics and structure of

the focus groups.

2.2 Topics and Interview Guideline

In alignment with a second team of researchers

responsible for conducting the quantitative web-

survey addressed above we developed a qualitative

and semi-structured discussion guideline which was

used by the moderators in all focus groups. This

guideline consisted of the following four main

constituents and topics.

2.2.1 The Perceived Importance of Mobile

Phones in Everyday Life

This part of the group discussion served as a warm-

up and addressed the importance of mobile phones

in everyday life situations of users as well as their

needs with regards to the security of applications

and the personal data stored on mobile phones or

smartphones. The usage of PINs was also discussed.

First, we wanted the participants to describe their

mobile phone usage, e.g. for which purposes,

business or private, and how often they use their

mobile phones. In a second step, they had to name

their most frequently used applications and the types

of data stored on their mobile devices. The

moderator noted the respondents’ reflections on big

post-its and collected them on a flipchart together

with a scale measuring the sensitivity of the

applications and data from 1 to 6. The value 1

represented “high sensitivity”, 6 “low or no

sensitivity”

. The participants then had to rank the

different mobile applications and data according to

this scale based on a group consensus. After the

ranking the users discussed their awareness of the

risks of insufficiently secure authentication methods

as well as the security level of PINs.

2.2.2 Alternatives to PIN-based

Authentication

The second part of our moderated group discussion

dealt with the participants’ experiences with

authentication methods alternative to the PIN. We

asked them to name all authentication methods they

had heard of or had already personally tried. The

participants’ reflections were again collected on a

flipchart.

2.2.3 Demonstration of Different

Authentication Methods

In the third part the participants had the opportunity

to evaluate different novel authentication methods

presented by the moderator. After testing each

method they had to discuss the benefits and

shortcomings of each one. We presented the

following eight different methods:

 Fingerprint Authentication

 3D Gesture Recognition

 Retina Scan

 Activity Based Verification

 2D Gesture Recognition

 Recognition Based Authentication

 Speaker Recognition

 Face Recognition

To minimize the risk of ranking effects we

varied the order in which the methods were

presented in the four different groups. Each

authentication method was presented on its own

“Very sensitive” meant that the users would very much want

to prevent unauthorized use of application and access to data

and “Not sensitive at all” meant that they would not mind

unauthorized use.

SECRYPT 2010 - International Conference on Security and Cryptography

158

along with typical usage scenarios. Afterwards, the

participants discussed the single methods in terms of

their subjective feeling of guaranteed security,

usability, convenience and intention of future usage.

After all eight authentication methods were

presented and discussed, the participants again

ranked them on the same 6-point scale. This time the

values represented the perception of the guaranteed

security provided by the methods. 1 meant that the

authentication method provided a “very high” and 6

a “very low” level of security.

2.2.4 Concluding Discussion

In the concluding part of our focus groups we then

re-evaluated the rankings on both scales and

compared them with each other. This gave us the

opportunity to directly match the perceived

sensitivity of applications with the expected security

of novel authentication methods and thereby gain

insights on whether a graded security system should

be provided together with different authentication

methods. The overall goal was to find out whether

some authentication methods were more suitable

than others for a graded security system and which

types of data fit best together with which type of

authentication method.

2.3 Demonstrator Showcases

At the time we conducted the focus groups, no

available technology existed on the market that

combined gradual security with alternative

authentication. We therefore had to use different

methodological approaches to demonstrate the

various knowledge- and biometry-based

mechanisms. This section describes the different

authentication mechanisms in more detail.

Fingerprint Authentication is authentication

based on the patterns of the human fingertip. We

presented this method using an integrated fingerprint

reader on a Lenovo notebook. In each group two

participants were requested to register into the

system with their fingerprints and then log in using

the fingerprint reader. After a successful login the

un-registered participants were asked to log into the

system as well but ultimately failed in doing so.

3D Gesture Recognition is a method with which

users can authenticate themselves through gestures

made in “free air” using the motion detectors

integrated into a broad variety of smartphones.

Because our 3D Gesture demonstrator was being

updated at the time, we used the analogy of

Nintendo’s Wii controller to explain the idea.

Retina or Iris Scan is a biometric mechanism

which authenticates users through scanning the

human eye using the mobile phone’s camera. In

order to explain this approach, pictures from a

technology fair presentation were used, as currently

no publicly available end device with such a

technology exists.

Face Recognition, a method we also discussed

with the respondents, is authentication based on the

recognition of the user’s face through the camera

integrated into the mobile phone. This concept was

explained together with Retina Scan authentication.

Activity Based Verification is based upon

password authentication. But in addition to a

password, the system also authenticates the user

according to the rhythm, speed and pressure of the

individual keystrokes. This method was presented on

a notebook. The technology itself was developed by

our engineers. In order to train the system the users

had to type in their passwords ten times. After that,

the registration was completed allowing the

participant to successfully log in. The other group

members who all knew the password were requested

to also log in but all of their attempts were to no

avail, mainly because the user-specific way of

stroking the keys could not be copied.

2D Gesture Recognition allows authentication

through a gesture drawn on the PC touchpad or

smartphone touchscreen. Authentication is achieved

through knowledge of the gesture itself but also

through the pressure on the touchpad and the speed

applied for “drawing” the gesture. This mechanism

was also presented with the help of a demonstrator

installed on a Lenovo notebook.

The Recognition Based Authentication method

works through selecting points on a picture in a

specific order as an alternative to the PIN with a

higher security level and better usability. The picture

can be chosen by the user and thus be a portrait of

family members or pets, for example. We

demonstrated this mechanism using an Android-

based application on the G1 smartphone, developed

by our engineers.

The final authentication method we presented in

our focus groups was Speaker Recognition. This

method is based on recognition of the speaker’s

voice using the microphone integrated into the

mobile phone. To visualize this biometric approach

we showed a video sequence of “A Space Odyssey”

where a man is requested to authenticate himself by

saying his name and destination before entering a

spaceship terminal.

It has to be taken into account that the evaluation

of a product or prototype can be influenced

"MY SMARTPHONE IS A SAFE!" - The User's Point of View Regarding Novel Authentication Methods and Gradual

Security Levels on Smartphones

159

depending on its respective quality at the time of the

users interaction with it. In our case we could not

notice a significant bias in the evaluation of the

security levels of each technology because the users

were generally able to abstract between the quality

of the technology at hand and the perceived future

benefits of it with regards to improved security.

2.4 Evaluation of the Data Obtained

All focus group sessions were recorded both on

audio and video using HD portable cameras and

MP3 voice recorders. In addition, a student assistant

protocolled each session. After all focus groups were

concluded, the interviewers as well as the keepers of

the minutes jointly evaluated both the video and

audio files and cross-referenced them with the

written notes to check whether there were variations

between video/ audio material, minutes and the

perceptions of the interviewers. After the findings

were written down in a research report, we conveyed

the results to the colleagues responsible for the

technical setup of the prototypes and demonstrators.

They re-evaluated our findings, commented and

added their impressions in order to finalize the

report. In addition, the respondents’ demographic

information surveyed at the end of the focus groups

through a three page questionnaire was evaluated by

a student worker using Microsoft Excel.

3 RESULTS AND FINDINGS

3.1 Focus Group Results

3.1.1 Evaluation of Authentication Methods

Our evaluation of the focus groups clearly showed

that users make distinctions between the perceived

security of the authentication methods presented,

their perceived convenience as well as the future

likelihood of use. Besides focussing on group

differences we also analysed whether gender-related

distinctions could be unveiled. After our test persons

were shown the different demonstrators described in

subsection 2.3, a group discussion was initiated by

the moderator, addressing the weaknesses and

benefits of each method. Adjacently, each

respondent individually evaluated the authentication

methods using a valence method based on smileys

The results were ranked by us, according to the

respective frequency of the users’ replies.

Table 2: Respondent agreement (ranked) to statement: “I

think this authentication method is secure”.

Table 3: Respondent agreement (ranked) to statement: “I

think this authentication method is good”.

Table 4: Respondent agreement (ranked) for statement: “I

think I would use this authentication method”.

One of the major insights we discovered was that the

perceived security of an authentication method does

not necessarily correlate with the overall acceptance

of the method or the willingness of use, as the tables

above show: The findings presented here highlight

that, with regards to the users’ perceived sensitivity,

Valence is an affective evaluation of approval/ disapproval, or

like/ dislike. The respondents could state whether they liked a

method, perceived it to be secure or whether they intended to

use it based on a “happy” and “sad” smiley.

SECRYPT 2010 - International Conference on Security and Cryptography

160

biometric authentication clearly outperforms all the

other methods presented. This was not only revealed

through the valence method, but also mentioned

manifold in the moderated discussion itself. Our

respondents perceived biometric authentication to be

the most secure

. For example, one user commented

that biometric authentication is “much easier

because you can’t forget it. You always have your

finger, eye or voice with you”. However, this did not

automatically mean that the same users also

perceived the method as good. Concerning Speaker

Recognition for example, it was stated that “it’s

strange to talk to the phone.” and “I wouldn’t use it

in public”. In addition, the perceived security does

not indicate a willingness of use, as a comparative

analysis of tables 2 and 4 shows. Even though all

users in the sample (100%) opined that Retina Scan

was the safest method (“pretty secure”), only 37%

liked the method as such and even less (26%)

actually stated that they would like to use it in the

future. Amongst others, reasons given by the users

were: “Embarrassing! It looks kind of strange, a bit

like James Bond…” and “Not feasible”. On the other

hand, methods that were perceived as secure by

approximately two thirds of the sample such as 2D

Gestures (63%) ranked second highest with regards

to acceptance (68%) and willingness of use (63%)

“It’s very intuitive”. The only (biometric)

authentication method that performed equally well

in all three categories was Fingerprint

Authentication with a perceived security of 95%, an

overall acceptance of 89% and a willingness of use

of 95%. For R&D departments and those dedicated

to developing future authentication methods or

security related applications, this is a remarkable

finding. With regards to gender related differences,

we discovered that male and female users had quite

different opinions on the presented authentication

methods, as is summarized in the following tables:

As our respondents mentioned one reason for the perceived

security of biometric and specifically Retina Scan

Authentication is the constant promotion of this security in

various science fiction movies.

Table 5: Gender related differences in overall acceptance

of authentication method.

Table 6: Gender related differences with regards to

perceived security of authentication method.

Table 7: Gender related differences with regards to future

willingness of use of authentication method.

The main finding regarding these different opinions

is that our participating women showed a

considerably greater affinity to new authentication

methods than the male ones, regarding the perceived

security, the perceived convenience as well as the

future likelihood of use: Apart from Speaker

Recognition, women perceived every method as

more secure than men. The same trend can be

identified regarding the perceived convenience:

Whereas seven out of eight methods were perceived

as good by at least 50% of the women, only three of

them were accepted by the men to that extent. These

"MY SMARTPHONE IS A SAFE!" - The User's Point of View Regarding Novel Authentication Methods and Gradual

Security Levels on Smartphones

161

findings result in the fact that for five out of eight

presented methods more than half of the women

stated that they would use it. Only two

authentication methods achieved a comparable

percentage in the men’s judgement. Particular

attention should be paid to the methods 3D Gestures

and Recognition Based Authentication which

explicitly revealed a great difference between

genders: Whereas only 18% respectively 40% of the

male participants intended to use these methods if

possible, nearly two thirds of the women quoted that

they could imagine using them. Similar numbers

hold for the perceived security and the overall

acceptance: At most one third of the men perceived

the respective methods as secure, whereas at least

50% of the women opined that they were secure.

The overall judgement also differed between

genders: At most 40% of the men liked these

methods, compared to at least 63% of the women

3.1.2 Evaluation of graded Security Concept

Besides evaluating the perceived security,

acceptance and usage intention of novel

authentication methods, our test persons also had to

discuss the concept of gradual authentication for

different types of data and applications. With the

help of the 6-point scale described in subsection 2.4

we asked our respondents to first list all the

applications they use and then discuss the respective

sensitivity of each application or data on a group

level. In most cases, the test persons in each specific

group mostly agreed on the same sensitivity level.

The differences between the groups however were

quite obvious. For business users emails were

considered to be a lot more sensitive than for

students and younger users. These users, on the

other hand, considered personal pictures or videos

stored on their devices to be more sensitive than

others and the group of parents considered

applications and data unsuitable for children to be

amongst the most sensitive types.

Based on the discussions within the different

groups we discovered that in general, the idea of

gradual security was perceived as useful by all test

With regards to the methods Retina Scan and Face

Recognition we also discovered that the majority of our

participating women would refrain from using these biometric

methods because of expected health concerns (“It hurts the

eyes”) whereas the male respondents mainly named usability

and feasibility aspects as the main counterarguments for a

future use of these methods.

persons. However, the individual requirements as to

how this system should work varied greatly,

revealing an overall need to customize the system

(“I’d really like to have the choice”). For example

the test users mentioned that they would like to

decide for themselves which data should be assigned

to which security level, rather than using previously

configured settings (“it should be possible to

configure the security settings individually right at

the beginning as you can do it with the internet

explorer, for example”).

Regarding group differences, the business users

specifically mentioned the need for authentication

methods with high security standards that

synchronistically have a low impact on the workflow

with the device (“it simply has to stay feasible”). For

the majority of the other “non-business” users,

usability and practicality were more important than

security aspects even after the awareness of possible

risks was increased through the group discussion.

The idea of having a gradual security system with

different authentication methods was favoured most

by parents because it allows sharing one smartphone

among family members without having to bother

about private data or cost-intensive applications (e.g.

a child pressing a button and calling an international

phone number, thereby increasing the phone bill).

The most prominent findings explicitly but also

implicitly stated by the users with regards to such a

graded security system were that [1] people should

feel more secure without constantly having to bother

with authentication. The authentication methods

therefore should have a high degree of usability. [2]

In general, it can be stated that the need for

improved security automatically arose with the

awareness of the risks of smartphone usage with

regards to data leakage, data theft or privacy threats

and that device manufacturers, OS developers or

MNOs should work on concepts to deal with these

increasing threats. [3] Since the way in which people

interact with their devices varies greatly over

different consumer segments, developers should

address this issue by providing pre-configured and

segment-specific solutions that could then be

customized further if the default settings do not

already fully meet the specific user needs. [4] A

final and useful remark was that there should always

be some kind of “backdoor” similar to the PIN/ PUK

approach, no matter how secure the other

authentication methods would be. To better explain

this remark, a focus group respondent gave the

examples of a sore throat or a blister on the fingertip,

interfering with the authentication on the device and

thereby rendering it “unusable”.

SECRYPT 2010 - International Conference on Security and Cryptography

162

3.2 Quantitative Cross-validation

In order to give our qualitative findings more

“weight” with regards to the perceived security of

the novel authentication methods, we had the

findings cross-validated through an international

web survey conducted by our colleagues based on

the same questions. The survey was in English and a

total of n=308 respondents participated. The gender

distribution across the quantitative sample was 45%

female and 55% male. 14% had separate phones for

business and private purposes. A sample distortion

led to the uneven distribution of approx. 75% Israeli

respondents and 25% German respondents.

Since we used closed answers for the

questionnaire, the results can not be compared to the

qualitative findings, which are based on “open

answers”. In addition, the respondents in our survey

were only confronted with written descriptions of

these authentication methods. This however was not

a problem per se, since we did not aim at obtaining

statistically representative and comparative results

but merely on capturing a glimpse of the “bigger

picture”. The 6-point Likert scale used in the web

survey shows similar tendencies regarding the

perceived security of smartphones. In the web

survey biometric authentication methods were also

perceived to be the most secure with Fingerprint

Authentication holding the pole position. Just like in

our focus groups, 2D and 3D authentication methods

were rated the least secure in comparison to the

other methods discussed in the web panel. The

following chart gives a comparative overview:

Figure 1: Security of authentication methods as perceived

by participants (n=308) of an international web survey.

The six possible answers were grouped together into

tendencies, with the two highest values, i.e. “6= very

secure” and “5= quite secure” grouped together to

“high security” as well as the two lowest values, “1=

not secure at all” and “2= not very secure”, being

grouped together to “low security”.

4 CONCLUSIONS

As an overall conclusion we can state that we

revealed useful insights for R&D personnel, even

though the sample size of our research was low and

the findings therefore not representative. In general,

the test persons agreed that gradual security is a

useful concept and that it would help to improve the

security of personal data and information stored on

smartphones or mobile phones. The test persons also

agreed that the PIN as the only standardised

authentication method does not sufficiently protect

the personal data stored. With regards to the overall

acceptance, intended usage and perceived security,

the findings show that out of all biometric

authentication methods, only Fingerprint

Authentication was fully accepted in all three

categories. The idea of combining a gradual security

system with different authentication methods

however did not meet with great approval. Merely a

few of the test persons favoured the idea of

combining low sensitivity data with lower security

mechanisms and high sensitivity data with very

secure mechanisms. The majority of the test persons

had the opinion that it was best to use high security

mechanisms for all types of data, in order to increase

the overall security of the device rather than using

different security levels. This position was supported

by the fact that the seemingly most secure method,

Fingerprint Authentication, was perceived as the

most usable as well.

As always the fact with qualitative data, results

have to be interpreted carefully. We were able to

obtain in-depth insights concerning our gradual

security approach as well as on the weaknesses and

strengths of novel authentication methods which we

would not have been able to obtain to the same

extent through a quantitative approach. With regards

to new and innovative ICT products and services,

qualitative approaches, implemented at an early

stage in the development process, are valuable as

they can give hints and directions to researchers and

engineers alike. They should then, at a later stage of

the process, be backed up with more “solid”

quantitative figures but also retested qualitatively in

an iterative development process. Our web-panel

with a total of n=308 participants gave additional

input but lacked statistical representativity. The

sample size was too low and unevenly distributed

among only two countries. The quantitative results

briefly discussed in this paper nonetheless show that

the opinions obtained in the focus groups were

mostly shared by users, who were not able to

actually test the demonstrators themselves. This

"MY SMARTPHONE IS A SAFE!" - The User's Point of View Regarding Novel Authentication Methods and Gradual

Security Levels on Smartphones

163

shows that the qualitative results can not merely be

attributed to the specific group characteristics but

possess a higher validity because they are shared by

larger user numbers.

For future research going into a similar direction

we would recommend, based on the experiences

gathered from our work, to further increase the

sample size and more strongly focus on cultural

differences regarding user requirements and needs. It

is also advisable, especially when technology is to

be properly evaluated, that user research experts

team up with engineers and developers, to get the

most out of research approaches of this kind: namely

user insights that are directly translated into

technical requirements and guidelines for

development.

The quality of the results also depends greatly on

the quality and reliability of the technologies or

demonstrators presented to the users. It would be

interesting to see how and to what extent opinions

might change with products ready or almost ready

for market launch. The same holds for the duration

and the intensity of the test: Whereas the participants

of our focus groups could not really experience the

different authentication methods over time but

merely test or imagine them, their judgement might

differ if those methods were already implemented on

a smartphone. For this reason, the development of

novel authentication methods has to be advanced.

With regards to our initial research hypothesis,

namely that smartphones should provide different

security layers (graded approach) combined with

different authentication methods, we can state that,

based on the findings discussed in this paper, this

does not seem to be the case for our respondents.

Both dimensions of our concept were considered

practical, useful and good on individual accounts. A

combination of both was considered to be

impractical. This leads to the further research

question of how different security layers can be

achieved without different authentication methods.

In the course of our group discussions we could

discover that even if there is a light awareness of

possible security threats most of the respondents did

not really think about them or their possible

consequences. However, it was remarkable that this

awareness grew during the discussion with other

users. This finding may indicate that there is a lack

of information concerning security threats among

users. In order to bring new security mechanisms for

smartphones forward, user education concerning

security threats also has to be taken into account.

ACKNOWLEDGEMENTS

Special thanks go to Mr. Niklas Kirschnik and Mr.

Hanul Sieger, both PhD candidates at Deutsche

Telekom Laboratories and Berlin Institute of

Technology for providing the demonstrators and

showcases during the focus group sessions and for

giving advice with regards to the interview

guideline. They were also our “troubleshooters”

during the focus groups, assuring that all

demonstrators operated smoothly at all times. In

addition, we would also like to thank Prof. Joachim

Meyer and Noam Ben-Asher from Ben-Gurion

University of the Negev, Israel for conducting the

quantitative survey referred to in sub-section 3.2 and

for providing us with the data and the chart used in

figure 1 to further back the insights from our focus

groups.

REFERENCES

Canalys, 2009. Smartphone market shows modest growth

in Q3, Canalys research release 2009/112. URL:

http://www.canalys.com/pr/2009/r2009112.htm, Last

checked: 2009/12/04.

Dilanchian, 2009. Smartphone statistics and links. URL:

http://www.dilanchian.com.au/index.php?option=com

_content&view=article&id=577:smartphone-statistics-

and-links&catid=23:ip&Itemid=114, Last checked:

2010/01/13.

Gartner, 2009. Gartner Says PC Vendors Eyeing Booming

Smartphone Market. http://www.gartner.com/it/

page.jsp?id=1215932, Last checked: 2010/01/22

International Organization for Standardisation ISO, 1998.

Ergonomic requirements for office work with visual

display terminals ISO 9241-11.

Krueger, R. A. & Casey, M. A., 2000. Focus Groups: A

practical guide for applied research, 3

Edition. Sage:

Thousand Oaks, California.

McAffee, 2009. Mobile Security Report 2009. URL:

http://www.mcafee.com/us/local_content/reports/mobi

le_security_report_2009.pdf. Last checked:

2010/01/13.

Reinhardt, A. b., Furnell, S. M. & Clarke, N. L., 2009.

From desktop to mobile: examining the security

experience. In: Computer & Security, pp. 130-137.

Stewart, D. W., Shamdasani, P. N. & Rook, D. W, 2007.

Focus Groups: Theory and practice. Sage: Thousand

Oaks, California.

SECRYPT 2010 - International Conference on Security and Cryptography

164