Download Novel approaches to biometric security with an emphasis on liveness and coercion detection PDF

TitleNovel approaches to biometric security with an emphasis on liveness and coercion detection
File Size1.6 MB
Total Pages226
Table of Contents
                            Table of contents
List of figures
List of tables
1 Beginning to consider
	1.1 Context
	1.2 Aims
	1.3 Objectives
	1.4 Original Contribution to Knowledge
	1.5 Thesis Structure
	1.6 Published Work
2 Methodologies
	2.1 Research Question
	2.2 Appropriateness of the Research Design
	2.3 Methodology
		2.3.1 Philosophy
		2.3.2 Grounded Theory
		2.3.3 Experimental
	2.4 Qualitative and Quantitative Data
		2.4.1 Target Audience
		2.4.2 Research Uses
		2.4.3 Research Forms
	2.5 Research Process
		2.5.1 Background Gathering
		2.5.2 Characteristic Classification
		2.5.3 Taxonomical Differences
		2.5.4 Coercion Detection Development
		2.5.5 Taxonomical Suitability
		2.5.6 Taxonomy Testing
		2.5.7 Algorithmic Development and Application
		2.5.8 Algorithm Analysis
	2.6 Ethical Considerations
	2.7 Research Risks
	2.8 Conclusion
3 Biometric Background
	3.1 Biometric Security
	3.2 Biometric Architecture
	3.3 Biometric Security and Privacy
		3.3.1 Intrinsic failure
		3.3.2 Adversary Attacks
	3.4 Liveness Detection
	3.5 Coercion Detection
		3.5.1 Development Background
		3.5.2 Coercion Detection Characteristics
		3.5.3 Coercion Characteristics
	3.6 Biometric Conclusion
4 Taxonomy Development and Application
	4.1 Liveness Detection Development
	4.2 Liveness Detection Categorisation
	4.3 Liveness Detection Analysis
		4.3.1 Universality
		4.3.2 Uniqueness
		4.3.3 Permanence
		4.3.4 Collectability
		4.3.5 Performance
		4.3.6 Acceptability
		4.3.7 Circumvention
	4.4 Coercion Detection Development
		4.4.1 Voluntary Techniques
		4.4.2 Involuntary
	4.5 Coercion detection categorisation development
		4.5.1 Universality
		4.5.2 Uniqueness
		4.5.3 Permanence
		4.5.4 Collectability
		4.5.5 Performance
		4.5.6 Acceptability
		4.5.7 Circumvention
	4.6 Conclusion
5 Testing and Evaluation
	5.1 Taxonomy Testing
		5.1.1 Liveness Application
		5.1.2 Coercion Application
	5.2 Algorithm Development
		5.2.1 Justification
		5.2.2 Algorithm components
	5.3 Taxonomy Evaluation
		5.3.1 Interval to Ratio Scale
		5.3.2 Data Clarification
		5.3.3 Hardware reliance
		5.3.4 Lack of mobile relevancy
		5.3.5 Acceptance
	5.4 Algorithm Evaluation
		5.4.1 Single Data Reliance
		5.4.2 Data Specificity
		5.4.3 Interface development
	5.5 Conclusion
6 Conclusion
	6.1 Future Work
	6.2 Submission Goals
Appendix A Research Risks
Appendix B Taxonomy and Algorithm Data
	B.1 Universality Data
	B.2 Permanence Data
	B.3 Collectability Data
	B.4 Performance Data
	B.5 Acceptability Data
	B.6 Circumvention Data
	B.7 Algorithm Examples
Document Text Contents
Page 1

Novel approaches to biometric security
with an emphasis on liveness and

coercion detection

Peter William Matthew

Department of Computing

Edge Hill University

This dissertation is submitted for the degree of

Doctor of Philosophy

January 2016

Page 113

4.3 Liveness Detection Analysis 95

Pf =
Th + A + SS + At


Fig. 4.4 Performance equation

ID Technique Th A Ss At T Outputs

1 Hippus dilation test 5 1 2 2 2 5
2 Skin conductivity test 1 1 2 4 4 2
3 3d face analysis 2 2 5 5 5 2.8
4 ECG 2 1 2 5 3 3.3

Table 4.10 Performance testing characteristics


This metric culminates in Figure 4.4 which identifies the overall level of performance a
liveness detection technique has.

Time is the most impactful value in this metric, that is why it is used to divide the rest of
the calculation, the quicker the time the higher the value and therefore the lower the overall
value. To test this equation the data in Table 4.10 will be considered:

Then the outputs in Table 4.11 will be calculated.
Table 4.11 shows that skin conductivity response test has the most performance and

performs twice as well compared to hippus dilation test mainly due to the technique hetero-
geneity. It also shows the ECG and 3d face analysis tests are very similar which is surprising,
as ECG is a a very new approach whilst 3d face analysis have been used for a number of
years and are well researched within the area.

4.3.6 Acceptability

The acceptance of new technology is potentially an uphill struggle as it is often the case
that technology, which is perfectly viable, will not be taken up due to an adverse public
opinion [83]. Therefore the acceptability of any liveness detection technique must be relevant
enough to allow a seamless integration, minimising any user acceptance problems. There
are legitimate concerns that users may refuse to use, or be unhappy about using, certain
techniques that they feel are invasive, unsafe, or insecure and whilst this can be alleviated
somewhat with a good knowledge of the devices and a transparent identification of data
flows there will always be some aversion to the technology. This is especially relevant when
considering biometric security as they are a comparatively new security technique compared
to pin numbers and passwords. Unfortunately the normal route the general public gets to hear

Page 114

96 Taxonomy Development and Application

Test Calculation

1. Pf =
5+ 1+ 2+ 2

2 = 5
2. Pf =

1+ 1+ 2+ 4
4 = 2

3. Pf =
2+ 2+ 5+ 5

5 = 3
4. Pf =

2+ 1+ 2+ 5
3 = 3.3

Table 4.11 Performance metric testing

about biometric devices is when they have been demonised in the media. This is done with a
variety of erroneous results popularised within media which often are designed to provide
entertainment without any thought regarding the realistic attributes of the technology.

Liveness detection has both advantages and disadvantages when considering acceptability
issues and one of the primary disadvantages is that the name, liveness detection, denotes a
medical emphasis which can potentially worry people who are unaware of its true purpose.
To alleviate this issue, liveness detection must endeavour to be as transparent as possible,
utilising as non invasive techniques as possible. By not providing the user with an outlet to
notice the sample collection process can improve the overall acceptance and integration of
the technology into daily routine. However there is an alternative view that this may produce
ethical and legal ramifications this technique would encounter. Whilst to some users the
transparency would make the technique more usable, others it would further alienate for fear
of what the data is being gathered for and why. An excellent example of this is the XBOX
Kinect, which was included into the games console the XBOX One, as it was identified that
the Kinect would constantly capture data to improve the overall service, however due to
colossal user backlash this factor was rescinded. Therefore if this technique transparency was
to be used it would have to be made clear to the users that there will be some form of liveness
detection gathering in progress, and that the data was only to be used for stated services.
This is only relevant when considering inherit and innate characteristics, as anything that
demands the user to provide additional data, or any technique that demands the user engage
with additional hardware after the authentication process, will automatically provide the user
with the knowledge of sample gathering.

It is almost impossible to identify people’s opinions without some form of primary data
collection, identifying what level of knowledge and acceptance a user base would have when
dealing with a specific technique. This would allow a level of generalisation to be identified
and therefore measured. Therefore to gather acceptance data a different technique must be
considered when compared to the other metrics. Whilst the acceptance of a technique is a
very subjective approach concerned mainly with qualitative data, it is a simple task to convert
this into a numerical measure by utilising a simple ranking system such as the Likert system.

Page 225

B.7 Algorithm Examples 207

B.7 Algorithm Examples

Page 226










sum t Time Elapsed (hours) Participants (ai/li) Time for Response Anomolous Userbase ai-Ab/ai Dr + K
1 1 5.010588235 2 1.875 0.995748299 4.162087384
3 2 6.010588235 1.64 2 0.995464853 4.162087384
5 3 7.010588235 1.65 1.625 0.996315193 4.162087384
7 4 8.010588235 1.470588235 1 0.997732426 4.162087384
9 5 9.010588235 2.25 1.25 0.997165533 4.162087384

Table B.14 Algorithm Example I

E*F Security Level - As Exp(ai-Ab/bi)/T lt ct ai-Ab/ai
4.144391434 1.795403646 1.353374521 1.513888889 1.373722222 0.995748299
4.143211704 1.202186181 1.649988983 1.513888889 1.373722222 0.995464853
4.146750894 1.013077277 1.641384191 1.513888889 1.373722222 0.996315193
4.152649544 0.776173992 1.844244937 1.513888889 1.373722222 0.997732426
4.150290084 1.055550236 1.204705714 1.513888889 1.373722222 0.997165533

Table B.15 Algorithm Example II

Similer Documents