Outline
Introduction
HISPs — Proposed OOB Methods
Experimental Design
Results
Analysis and Discussion
Usability and Security of Out-Of-Band
Channels in Secure Device Pairing Protocols
Ronald Kainda, Ivan Flechais, and A.W. Roscoe
Oxford University Computing Laboratory
SOUPS Conference
15-17 July, 2009
Conclusion
Outline
Introduction
HISPs — Proposed OOB Methods
Experimental Design
Outline
1
Introduction
2
HISPs — Proposed OOB Methods
3
Experimental Design
4
Results
5
Analysis and Discussion
6
Conclusion
Results
Analysis and Discussion
Conclusion
Outline
Introduction
HISPs — Proposed OOB Methods
Experimental Design
Outline
1
Introduction
2
HISPs — Proposed OOB Methods
3
Experimental Design
4
Results
5
Analysis and Discussion
6
Conclusion
Results
Analysis and Discussion
Conclusion
Outline
Introduction
HISPs — Proposed OOB Methods
Experimental Design
Introduction - Device Pairing
Scenario
Results
Analysis and Discussion
Conclusion
Outline
Introduction
HISPs — Proposed OOB Methods
Experimental Design
Introduction - Device Pairing
Human-Interactive Security Protocols (HISPs)
OOB
N
Results
Analysis and Discussion
Conclusion
Outline
Introduction
HISPs — Proposed OOB Methods
Experimental Design
Results
Introduction - HISPs
Security in HISPs
Technical security
Security based on formal proofs
Depends on the size of the digest/fingerprint
b-bits for most protocols
Analysis and Discussion
Conclusion
Outline
Introduction
HISPs — Proposed OOB Methods
Experimental Design
Results
Analysis and Discussion
Conclusion
Introduction - HISPs
Security in HISPs
Technical security
Security based on formal proofs
Depends on the size of the digest/fingerprint
b-bits for most protocols
Effective security
Secure systems are socio-technical (Sasse et al.)
Security of a protocol may depend on human effort
Humans forget, make mistakes
These mistakes may result in security failures
Human failures are not covered by formal proofs
Increasing technical security (value of b) may reduce effective
security
Outline
Introduction
HISPs — Proposed OOB Methods
Experimental Design
Results
Analysis and Discussion
Introduction - Research Question
Are proposed OOB methods usably secure to guarantee
specified technical security?
Conclusion
Outline
Introduction
HISPs — Proposed OOB Methods
Experimental Design
Outline
1
Introduction
2
HISPs — Proposed OOB Methods
3
Experimental Design
4
Results
5
Analysis and Discussion
6
Conclusion
Results
Analysis and Discussion
Conclusion
Outline
Introduction
HISPs — Proposed OOB Methods
Experimental Design
HISPs - Proposed OOB Methods
Manual comparison
compare
Devices generate
fingerprints
Fingerprints displayed in
appropriate format
Users compare fingerprints
and indicate on the device
a match or lack of it
Devices require display
and some form of input
method
Results
Analysis and Discussion
Conclusion
Outline
Introduction
HISPs — Proposed OOB Methods
Experimental Design
Results
HISPs - Proposed OOB Methods
Manual copying and entering
Bluetooth
One device displays a
fingerprint
User copies and types the
fingerprint into one or more
devices
Requires display and keypad
Efficiency of entry depends
on affordances of devices
involved
Analysis and Discussion
Conclusion
Outline
Introduction
HISPs — Proposed OOB Methods
Experimental Design
Results
Analysis and Discussion
HISPs - Proposed OOB Methods
Auxiliary devices
2D–Barcode
Rely on secondary devices to
transfer/compare information
Proposed devices include
camera phone
external storage devices
data cable etc
May require users to carry
extra hardware
Conclusion
Outline
Introduction
HISPs — Proposed OOB Methods
Experimental Design
Results
HISPs - Proposed OOB Methods
Short-range wireless channels
infrared
Rely on short range wireless
channels
Require devices to be no
more than a few centimetres
apart
Proposed methods include:
infra-red
light
distance bounding1
Most methods lack human
verification
1
can also use normal channel
Analysis and Discussion
Conclusion
Outline
Introduction
HISPs — Proposed OOB Methods
Experimental Design
Results
Analysis and Discussion
HISPs - Proposed OOB Methods
Timing methods
Rely on transmission of information in well timed intervals
Users coordinate the synchronisation
Examples include
shaking devices (Saxena et al.)
pressing a button in response to some stimulus
Conclusion
Outline
Introduction
HISPs — Proposed OOB Methods
Experimental Design
Outline
1
Introduction
2
HISPs — Proposed OOB Methods
3
Experimental Design
4
Results
5
Analysis and Discussion
6
Conclusion
Results
Analysis and Discussion
Conclusion
Outline
Introduction
HISPs — Proposed OOB Methods
Experimental Design
Results
Analysis and Discussion
Conclusion
Experimental Design - Methods
DEFINITIONS
Method
– refers to a specific mode of comparing/transferring information
between devices by humans
Representation
– refers to specific format in which information is presented to users
Method-representation
– refers to a combination of a method and representation
Outline
Introduction
HISPs — Proposed OOB Methods
Experimental Design
Results
Analysis and Discussion
Experimental Design - Method-representations
Compare & confirm
compare
Numeric
Alphanumeric
Words
Sentences
Country names
Numeric & Sound
Alphanumeric & Sound
Melodies
Images
Conclusion
Outline
Introduction
HISPs — Proposed OOB Methods
Experimental Design
Results
Analysis and Discussion
Experimental Design - Method-representations
Compare & select
Numeric
Alphanumeric
Copy & enter
Numeric
Alphanumeric
Barcode
Conclusion
Outline
Introduction
HISPs — Proposed OOB Methods
Experimental Design
Experimental Design - Design
Dependent variables
1
Time
2
Number of non-security failures
3
Number of security failures
Independent variable
1
Method-representation
Results
Analysis and Discussion
Conclusion
Outline
Introduction
HISPs — Proposed OOB Methods
Experimental Design
Results
Analysis and Discussion
Experimental Design - Participants
30 participants were recruited via online advertisement
Gender
Age
Education
Male: 47%
Female: 53%
18 - 25 40%
26 - 35 27%
36 - 45 13%
46 - 55 3%
56 - 65 13%
66 - 75 4%
High School: 27%
College: 27%
Graduate: 26%
Postgraduate: 20%
Conclusion
Outline
Introduction
HISPs — Proposed OOB Methods
Experimental Design
Results
Experimental Design - Apparatus
tools
Devices: Nokia N95 and N73
Nokia devices are common
Bluetooth enabled
Software:
P2P payment system
Device communication using
Bluetooth
Software created a log of
participant’s actions
Digital voice recorder
To record interviews
Questionnaires
Enrolment
After scenario (AS)
After experiment/exit (AE)
Written instructions
Analysis and Discussion
Conclusion
Outline
Introduction
HISPs — Proposed OOB Methods
Experimental Design
Results
Experimental Design - Procedure: Tasks
Figure: Step 1
Analysis and Discussion
Conclusion
Outline
Introduction
HISPs — Proposed OOB Methods
Experimental Design
Results
Experimental Design - Procedure: Tasks
Figure: Step 2
Analysis and Discussion
Conclusion
Outline
Introduction
HISPs — Proposed OOB Methods
Experimental Design
Results
Experimental Design–Procedure: Tasks
Figure: Step 3 and 4
Analysis and Discussion
Conclusion
Outline
Introduction
HISPs — Proposed OOB Methods
Experimental Design
Results
Experimental Design - Procedure: Tasks
Figure: Step 5 and 6
Analysis and Discussion
Conclusion
Outline
Introduction
HISPs — Proposed OOB Methods
Experimental Design
Outline
1
Introduction
2
HISPs — Proposed OOB Methods
3
Experimental Design
4
Results
5
Analysis and Discussion
6
Conclusion
Results
Analysis and Discussion
Conclusion
Outline
Introduction
HISPs — Proposed OOB Methods
Experimental Design
Results
Analysis and Discussion
Conclusion
Results - Compare & confirm: Errors and completion times
Numeric
Alphanumeric
Words
Images
Country/
City names
Sentences
Alphanumeric
& sound
Numeric &
sound
Melodies
Time (s)
Mean
6
6
7
8
9
Security failures
%
0
13.3
3.3
0
0
Non-security failures
%
3.3
16.7
16.7
3.3
3.3
11
12
0
3.3
16.7
20
14
3.3
0
24
6.7
36.7
Between-subjects: p = 0.0007
Within-subjects - time: p = .0000
Outline
Introduction
HISPs — Proposed OOB Methods
Experimental Design
Results
Analysis and Discussion
Conclusion
Results - Compare & select: Errors and completion times
Numeric
Alphanumeric
Time
Seconds
9
9
Security failures
%
10
20
Between-subjects: p = 0.0000
Within-subjects - time: p = 0.9255
Non-security failures
%
10
30
Outline
Introduction
HISPs — Proposed OOB Methods
Experimental Design
Results
Analysis and Discussion
Results - Copy & enter: Errors and completion times
Numeric
Alphanumeric
Time
Seconds
17
40
Between-subjects: p = .7531
Within-subjects - time: p = .0004
Non-security failures
%
13
23
Conclusion
Outline
Introduction
HISPs — Proposed OOB Methods
Experimental Design
Results
Analysis and Discussion
Results - Barcode: Errors and completion times
Time
Seconds
37
Non-security failures
%
53
Conclusion
(C
C)
(C
ic
)
E)
(C
(a) Difficult
er
ic
E)
E)
(C
(C
en
um
nt
Se
ic
(C
er C)
ic
(C
C)
ce
Im s(C
C
ag
es )
W (CC
or
ds )
M
(
el
od CC
)
N
um Co ies(
A
un
lp
CC
ha eri
t
)
nu c & rie
s(
m
C
S
er
ic oun C)
d(
&
C
So
un C)
N
um d(C
A
lp
ha eri C)
nu c (
CS
m
er
ic )
(C
Ba S)
rc
od
e
er
um
um
0
N
10
an
60
lp
h
% Participants
70
A
ic
20
er
30
um
40
N
Difficult
Avoid
an
50
lp
h
% participants
HISPs — Proposed OOB Methods
A
ce
C)
Im s(C
C
ag
es )
W (CC
or
)
d
M
el s(C
od
C)
N
um Co ies(
A
un
lp
CC
ha eri
t
)
nu c & rie
s(
m
C
S
er
ic oun C)
d(
&
CC
So
un
)
N
um d(C
A
lp
ha eri C)
nu c (
C
m
S
er
ic )
(C
Ba S)
rc
od
e
en
er
m
ic
ic
er
(C
E
Introduction
nt
Se
er
m
nu
ic
er
um
N
ha
lp
A
nu
um
N
ha
lp
A
Outline
Experimental Design
Results
Analysis and Discussion
Results - Preferences
50
45
40
35
30
25
20
15
10
5
0
Easy
Preferred
(b) Easy
Conclusion
Outline
Introduction
HISPs — Proposed OOB Methods
Experimental Design
Outline
1
Introduction
2
HISPs — Proposed OOB Methods
3
Experimental Design
4
Results
5
Analysis and Discussion
6
Conclusion
Results
Analysis and Discussion
Conclusion
Outline
Introduction
HISPs — Proposed OOB Methods
Experimental Design
Results
Analysis and Discussion
Analysis and Discussion – SUM Score Ranking
Conclusion
Outline
Introduction
HISPs — Proposed OOB Methods
Experimental Design
Results
Analysis and Discussion
Conclusion
Analysis and Discussion – Ranking by Security Failures
Numeric(CE)
Alphanumeric(CE)
Barcode
Numeric(CC)
Sentences(CC)
Countries(CC)
Images(CC)
Words(CC)
Numeric & sound
Alphanumeric & sound
Melodies(CC)
Numeric(CS)
Alphanumeric(CC)
Alphanumeric(CS)
Subject to SF
No
No
No
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
SF
0
0
0
0
0
0
0
3.3
3.3
3.3
6.7
10
13.3
20
SUM Score
69
60.4
53
73.7
62.9
59.1
54.3
70.6
69.2
65.8
40.7
68.3
72.5
64.2
Outline
Introduction
HISPs — Proposed OOB Methods
Experimental Design
Results
Analysis and Discussion
Conclusion
Discussion and Discussion – Security Vs Usability trade-off
Outline
Introduction
HISPs — Proposed OOB Methods
Experimental Design
Results
Analysis and Discussion
Discussion – Security Vs Usability Considerations
User conditioning
User motivation
Security failures
Attentiveness
Conclusion
Outline
Introduction
HISPs — Proposed OOB Methods
Experimental Design
Outline
1
Introduction
2
HISPs — Proposed OOB Methods
3
Experimental Design
4
Results
5
Analysis and Discussion
6
Conclusion
Results
Analysis and Discussion
Conclusion
Outline
Introduction
HISPs — Proposed OOB Methods
Experimental Design
Results
Analysis and Discussion
Conclusion
Traditional methods are favoured by users
Currently proposed methods need rethinking about their
security/usability
Security failures are not acceptable
To achieve human compliance, enforcement is required
Copy & enter is the best compromise
Conclusion
Outline
Introduction
HISPs — Proposed OOB Methods
Experimental Design
THANK YOU
Results
Analysis and Discussion
Conclusion
Outline
Introduction
Conclusion
HISPs — Proposed OOB Methods
Experimental Design
Results
Analysis and Discussion
Conclusion
Outline
Introduction
Conclusion
HISPs — Proposed OOB Methods
Experimental Design
Results
Analysis and Discussion
Conclusion