Evaluating and selecting the biometrics in network security

RESEARCH ARTICLE

Evaluating and selecting the biometrics in network

security

Che-Hung Liu

*, Jen-Sheng Wang

_{, Chih-Chiang Peng}

_{and Joseph Z. Shyu}

2 _{Institute of Technology Management, National Chiao Tung University, Hsinchu, Taiwan}

ABSTRACT

Since Apple merged with AuthenTec, a leadingfingerprint recognition company, in 2012, biometrics has widely been con-sidered to strengthen security and privacy in the network securityfield. Although biometrics has been applied in specific areas for decades, it has gradually proliferated in customer and mobile electronic products to enhance security and privacy. This study aims to evaluate biometrics through conventional technology assessment considerations combined with view-points on the specifics of biometric technologies and then to provide suggestions for selection. To conduct the biometric technology assessment, the fuzzy analytic hierarchy process and non-fuzzy best performance approaches are used. Although the outcomesfirst indicate that technology assessment should be the key object in selecting biometric technolo-gies, that object is followed by biometric competence and key elements of biometrics. The outcomes also indicate that fea-tures of the target technologies should be considered when evaluating them. Additionally, fingerprint recognition, iris recognition, and face recognition are the preferred biometrics in evaluation and selection. Copyright © 2014 John Wiley & Sons, Ltd.

KEYWORDS

biometrics; network security; fuzzy analytic hierarchy process; best non-fuzzy performance

*Correspondence

Che-Hung Liu, Department of Business and Management, National University of Tainan, Tainan, Taiwan. E-mail: chehung@mail.nutn.edu.tw

1. INTRODUCTION

Security and privacy have become major concerns in ev-eryone’s daily networking life. To face future challenges, many companies and institutions are devoting themselves to researching and developing biometric products, such as Apple’s purchase of AuthenTec in 2012 and launch of the iPhone 5S withfingerprint authentication [1,2]. Gener-ally, the main purpose of biometrics is to improve security through recognition of unique features of human bodies [3,4]. Biometrics should offer dependable and robust personal recognition for confirming or determining an indi-vidual identity [5]. Biometric applications are found across the network security field as part of such applications as secure electronic banking, computer systems security, mobile phones, secure access to buildings, credit cards, and social and health services [6,7].

However, a closer look reveals that the various areas that biometrics is anticipated to support all focus on its verification aspects [8]. In addition, despite the many eval-uations of biometric technologies, opinions are widely divided, with analyses coming mostly from the technology

side, lacking either management or marketing points of view [9]. In contrast, there are several technology management studies that have tried to address generic technology assess-ment models to evaluate certain technologies [10–12]. However, biometric technologies in network security are more distinctive and complicated. Hence, this research aimed to assess various biometric technologies applied in network security, a subject that ranked in the top six in the International Biometric Group’s investigation [13]. To com-bine the two major assessment perspectives of technology specialization and management, this study built and tailored a specific evaluation and selection model to understand the purposes of biometric technologies applied in network secu-rity. On the basis of the research results, we provide sugges-tions for relevant persons to consider.

Usually, technology assessment studies apply many so-phisticated analytical methods [10]. The analytic hierarchy process (AHP) built by Saaty is one of the most widely used [14]. It is worth noting that decision makers’ judg-ments about the problems that they want to solve are often ambiguous and uncertain. Fuzzy AHP (FAHP), which is AHP integrated with fuzzy set theory, could measure the Published online 3 July 2014 in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/sec.1020

ambiguity and uncertainty that exist in the subjective opinions of decision makers. In the study, we adopted FAHP to form a biometrics evaluation model for identify-ing and weighidentify-ing the criteria critical to the topic of this study.

The arrangement of this paper is as follows. Section 2 briefly introduces six biometric technologies. An evaluat-ing framework is constructed in Section 3 based on technology assessment theories and the specific character-istics of biometric technologies. Section 4 introduces the applied FAHP method. The research analysis conducted using FAHP appears in Section 5. Section 6 draws conclu-sions from the research summary results in Section 5 and discusses management implications.

2. LITERATURE REVIEW

We introduce six top-ranked biometric technologies from the IBG 2009 market report [13]. The definitions most likely are not exhaustive but are representative of the target biometrics in terms of technological maturity, capability, and potential applicability.

2.1. Biometrics

On the basis of the goal of this study, the definition of biometrics offered by the US Department of Defense’s Biometrics Identity Management Agency is more than suf-ficient to convey the two common meanings of the term [15]:“A general term used alternatively to describe a char-acteristic or a process. As a charchar-acteristic: The measure of a biological (anatomical and physiological) and/or behav-ioral biometric characteristic that can be used for auto-mated recognition. As a process: Autoauto-mated methods of recognizing an individual based on the measure of biolog-ical (anatombiolog-ical and physiologbiolog-ical) and/or behavioral biometric characteristics” [15,16]. Biometrics generally comprise a set of elements [17]:

• Biometric data-collecting components. • Biometric data-storing components. • Biometric data-processing components.

• Decision-making components regarding matches between biometric data and verification results. • Transmission components for aiding the data

collec-tion, data storage, and signal-processing components to compress and expand necessaryfiles in the differ-ent process stages.

Practical biometric systems should confirm the accu-racy, speed, and resource requirements for the specified recognition. They must also be harmless to the users, be accepted by the intended population, and be sufficiently robust against various fraudulent methods and attacks to the systems [16,18].

2.2. Face recognition

Face recognition is well known to humans and is the most natural biometric identification [19]. Human performance, however, declines with fatigue and repetition (e.g., an opera-tor verifying photo identification of passengers boarding an airplane) [15]. Face recognition can serve in standoff or covert biometric systems and can be combined with other biometrics to increase confidence in results [20]. With iris recognition, it is a strong contender for second place (in mar-ket share) among biometric systems. These biometrics can be the object of a variety of modalities, including 2D and 3D imaging, 2D/3D combination, thermograms, and various analytical methods for recognition [21]. Because of changes in facial appearance over time, this biometric generally requires periodic re-enrollment. It is the least intrusive of biometrics, but when combined with extensive surveillance camera systems, it can raise issues of privacy [22]. 2.3. Fingerprint recognition

Fingerprint identification is the leading biometric in terms of market share and is the oldest with a scientific record [23]. It is characterized by relative ease of enrollment and low error rates [24]. Recognition accuracy can be increased by using prints from multiple fingers and can easily be used in the field for forensic purposes [20]. Drawbacks include the need for contact with a sensor, degraded perfor-mance when in the presence of dirt or degraded finger-prints (by age, manual work, or injury), and requirements for intensive computation when trying to match a sample to templates in a large database. Modern approaches use livefingerprint readers based on ultrasonic, optical, silicon, or thermal principles [25,26].

2.4. Iris recognition

Automated iris recognition emerged in the 1990s. It sam-ples the iris of the eye, which is the colored area surround-ing the pupil. Iris patterns are unique and can be obtained with a video-based image acquisition system [7]. Iris rec-ognition relies on light to sense the unique features of a person’s iris [27]. These features could be a composition of specific traits comprehended as crypts, corona, freckles, filaments, furrows, pits, rings, and striations [20]. Iris rec-ognition has demonstrated low error rates in tests, and per-formance infield systems is improving [21]. It requires cooperation for enrollment, and enrollment may be re-quired over a lifetime. However, designers of an iris recog-nition system must take care to consider two influences: lighting conditions and the size of the iris change. Before computing the iris code, the system has to process a proper transformation [24,28].

2.5. Speaker recognition

This recognition identifies subjects using the temporal and spectral characteristic of an individual’s voice [20]. It uses

the acoustic characteristics of speech that differ from one individual to another. These acoustic patterns cover both learned behavioral patterns (e.g., speaking style, voice pitch) and anatomy (e.g., shape and size of the mouth and throat) [7,29]. Low to medium error rates are obtained, depending on the quality of the communication link and ambient noise, and can be affected by the speaker’s condi-tion (e.g., emocondi-tional state, health) [23]. The strength of speaker recognition is that it is, with qualified hardware, currently the only biometrics applicable to voice communi-cation systems over long distances [16].

2.6. Vascular pattern recognition

Vascular pattern recognition, also generally known as vein pattern authentication, is an emerging biometric compared with existing systems. This biometric relies on the unique pattern of blood vessels of an individual, generally using the back of the hand [20]. Using near-infrared light, it detects the transmitted or reflected images of the blood ves-sels of a finger, palm, or hand for personal recognition. This requires proximity to, but no contact with, a sensor (in contrast tofingerprints) [4]. Different vendors employ different parts of the fingers, palms, or hands but use a similar methodology. Researchers have confirmed that the vascular pattern of the human body is unique to a cer-tain person and does not change with age [30]. Vascular pattern recognition has been applied in various cases; it is quite popular in Japan for ATM and banking access [21]. 2.7. Palm print recognition

The inner surface of the palm commonly has threeflexion creases, secondary creases, and ridges. Theflexion creases are also named principal lines, and the secondary creases are named wrinkles. Even identical twins have different palm prints [31]. Palm print recognition inherently

employs many of the same matching traits that allowed fingerprint recognition to be one of the most famous and publicized biometrics. However, automation of palm recognition has lagged because of some constraints in live-scan technologies and computing capabilities [20]. The weaknesses of this biometric are lack of accuracy, size of the scanner, cost (the scanners are fairly expensive), and the fact that injuries to palms can hinder the system’s func-tion [4]. It has gained a niche market in the areas of access control and time/attendance monitoring, possibly because of the size of the sensor making it more practical forfixed applications [22].

3. THE EVALUATION FRAMEWORK

Technology assessment involves different perspectives of diverse stakeholders, including practitioners, decision makers, researchers, and R&D personnel in private and public sectors [10]. In general, concerns in technology assessment comprise technological, economic, technology development, and risk aspects [11]. Hence, the perspec-tives of these technology assessment methodologies should be taken into consideration as well.

This study constructs a tailor-made technology assess-ment framework for biometrics (Figure 1) following related literature and includes in-depth discussions with enterprises and experts in the biometrics field to ensure the validity of the proposed framework. The content of the objects in the analysis model and corresponding criteria are illustrated as follows.

3.1. Technology assessment

The considerations of technology assessment can be syn-thesized and distinguished into several criteria, which should be grouped into positive prospects and negative

problems existing in technology development by interpreting related literature and secondary documents to deeply analyze potential effects [11].

3.1.1. Technological merit.

Studies indicate that the technological aspect plays an im-portant role in technology assessment [32,33]. Technological merit can be viewed from the following aspects [11,32–34]: (1) advancement; (2) innovation; (3) key technology; (4) proprietary aspects; (5) generic aspects; (6) technological connections; and (7) technological extendibility.

3.1.2. Business effects.

Effects benefitting corporations and economic/indus-trial development are of considerable weight in the evalua-tion of technology. Hence, we should consider business factors. These include the following: (1) possible invest-ment returns; (2) existing market share variation; (3) new market growth; (4) possible scale of the market; and (5) product life cycle [11,32,34–36].

3.1.3. Technology development potential.

It is necessary to consider the availability of related technological resources upon technology assessment. Factors affecting the realization of technology develop-ment are as follows: (1) technical resource availability; (2) equipment support; and (3) opportunity for technical success [11,32,34,37].

3.1.4. Risk.

When assessing new technologies, decision makers are faced with potential risks associated with the technology development. The criteria of risk aspects are as follows: (1) commercial risk; (2) technical risk; (3) technical dif fi-culties; and (4) ethical risk [11,34,35,38].

3.2. Biometrics performance

Jain et al. grouped fundamental required performance in biometrics into four categories: (1) accuracy; (2) scale; (3) security; and (4) privacy [16].

3.2.1. Accuracy.

An ideal biometrics system should promise to offer the correct decision when presented with a biometric identifier sample hand [39]. Even ignoring the requirements of com-plete automation and assuming the possibility of good bio-metric signal acquisition from a distance, this need should clearly be noted when attempting to bridge the gap between performance requirements and current technology [16]. 3.2.2. Scale.

Because verification systems essentially involve a 1:1 match, the size of the database is not so critical, requiring only comparing one set of enrolled samples to one set of enrollment templates [39]. Efficient scaling is required for system control throughput and false-match error rates as the size of the database increases [16].

3.2.3. Security.

In the past, two serious criticisms of biometrics have been that they are not secret and that biometric patterns have not been appropriately assessed [40]. However, bio-metric template protection is a rapidly developing area; extensive research in this area has been conducted in the last several years to enhance security [41]. For example, cancellable templates are perfect examples to address the issue of biometric non-revocability [42]. A secure biomet-ric system is a design challenge requiring the ability to not be fooled by doctored or spoofed measurements injected into the system while accepting only the legitimate presen-tation of biometric identifiers [16].

3.2.4. Privacy.

A biometric system should provide irrefutable proof of the identity of the person with extreme reliability. In conse-quence, privacy is one of the most significant user con-cerns [16]. Much hard work is required to provide satisfactory solutions for this fundamental privacy prob-lem. There are several ingredients needed for a successful strategy [43].

3.3. Key elements of biometrics

There arefive common elements to all biometric systems. 3.3.1. Enrollment.

Proper enrollment instruction and training are essential to good biometric system performance [17]. Enrollment is thefirst stage for biometric system setup because it gen-erates the template that will be used for all subsequent comparisons and user recognition [15]. During enrollment, a biometric system averages readings or selects the best quality sample to produce an enrollment reference or template [18].

3.3.2. Biometric template (or reference).

The biometric system software will use a proprietary algorithm to extract features that are appropriate to that biometric as a template or reference [15,44]. Templates are usually not actual images of thefingerprint, iris, hand, and so on [17] but are instead generally only numerical representations of key data points (or minutia) read from a person’s biometric feature [18].

3.3.3. Comparison and comparison errors. Comparison is the act of comparing one (or more) ac-quired biometric samples to one (or more) stored biometric templates for the purpose of recognition [15]. No biometric decision is 100% perfect in either verification or identifica-tion mode [18]. Therefore, biometric systems can be con-figured to create a threshold establishing the acceptable degree of similarity [17].

3.3.4. Networking.

Regarding networks, there are potential variation issues. Biometric systems/readers require integral networking

functionality with a proprietary protocol [17]. This allows networking a number of readers together with little or no additional equipment. A monitoring PC typically connects at an endpoint of the network [15,18].

3.3.5. Personal biometric characteristics.

Any human biological or behavioral feature can become a biometric identifier, provided the following properties are met [16,18]: (1) universality: almost every person should have the characteristics; (2) distinctiveness: no two people should have identical biometric characteristics; (3) perma-nence: the characteristics should not vary or change with time; and (4) collectability: obtaining and measuring the biometric feature(s) should be easy, non-intrusive, reliable, and robust.

4. RESEARCH METHODS

This study mainly applied the FAHP to assess the feasibil-ity of using biometrics to achieve the evaluation objects and criteria. For determinations, it is critical not only to evaluate the priority of different objects but also to deter-mine whether the biometrics fulfills the implementation objectives. Biometric technology evaluations should be well defined to allow resources and elements to be allo-cated to achieve the objectives. In this case, AHP is often a quite-useful approach that encourages decision makers to clearly describe their subjective and qualitative judg-ments [14]. However, to enhance AHP result reliability, we adopt FAHP rather than conventional AHP. Further-more, the outcomes generated by FAHP complement the biometric technology preference recommendations pro-duced by best non-fuzzy performance (BNP) analysis. We illustrated the FAHP steps in this section.

The AHP can support the process of assessing the fitness rank of a group of factors and the relative priority of a multi-criteria decision-making issue between them [14,45,46]. At the same time, academics and professionals have extensively applied AHP in thefield of technology assessment [47–51]. In this biometric technology analysis, AHP permits the“hierarchization” of distinctive evaluation objects and their allied or related criteria, generating possi-ble quantitative results that deliver a numerical estimate of the relevant consequences of each criterion and alternative. Regardless, some researchers criticize AHP, claiming that it cannot accurately reflect human cognition because of the uncertainty and ambiguity of decision makers’ judg-ments [52–55]. It is difficult to understand the partiality of decision makers for certain numbers. The fuzzy method is proposed to improve the AHP [54,56,57]. Fuzzy set theory was introduced to solve puzzles with less-clearly defined criteria [58]. FAHP was developed to solve hierarchical problems. Decision makers are usually more confident giv-ing interval judgments thanfixed value judgments because they are usually unable to be explicit about their prefer-ences concerning the fuzzy nature of the comparison pro-cess [59]. Linguistic variables are variables whose values

are words or sentences in a natural or artificial language. In other words, they are variables with lingual expression as their values [54,57]. Hence, we use this type of expression to compare two approaches to creating the best plan evalua-tion criteria, applyingfive basic linguistic principles to a fuzzy level scale [60,61]. A similar structure concept com-bining FAHP with BNP has been adopted in many studies [62–64]. We feel that the scheme could be useful in this biometrics assessment study. The steps are as follows.

Step 1: Developing the assessment matrix structure Construct pairwise comparison matrices among all the elements/criteria in the dimensions of the hierarchy system. Assign linguistic terms to the pairwise comparisons by ask-ing which is more important for each two criteria for each of the K decision makers based on a nine-point scale [14].

Step 2: Checking for consistency

The AHP is applied to obtain the weights of the criteria, w1,…, wn, based on the hierarchy built in step 1. The re-ciprocal matrix A is a pairwise comparison, in which aijis

the geometric mean of the criteria i and j comparison.

A¼ aij
 ¼ 1 a12 ⋯ a1n a21 1 ⋯ a2n ⋮ ⋮ ⋱ ⋮ an1 an2 ⋯ 1 2 6 6 6 4 3 7 7 7 5 ¼ 1 a12 ⋯ a1n 1=a12 1 ⋯ a2n ⋮ ⋮ ⋱ ⋮ 1=a1n 1=a2n ⋯ 1 2 6 6 6 4 3 7 7 7 5 (1)

The priority of the elements was compared by the compu-tation of eigenvalues and eigenvectors in Equation (2), in which the eigenvector of the matrix A is w, which could be calculated per Equation (3). The largest eigenvalue of the matrixA is λmax, which could be the means that could be

es-timated per Equation (4), and n is the criterion number. Aw ¼ λmaxw (2) w¼ ∏ n j¼1aij !1=n

=

The consistency property of the matrix is then checked to ensure the consistency of the judgments in the pairwise com-parison. The consistency index (CI) and consistency ratio (CR) are defined as follows [14]. Random index (RI) in Equation (6) can be referred to in Table I according to criterion n.

CI¼ λmax n

n 1 (5)

CR¼CI

RI (6)

Step 3: Constructing fuzzy positive matrices To calculate criteria and sub-criteria weights to compare with each other, the Buckley method is used [65]. With the triangular fuzzy numbers in Table II, for calculating weights in each pairwise comparison matrix, the geometri-cal mean of each criterion was used in Equation (7), in which eAk is decision maker k’s fuzzy positive reciprocal matrix andeakijrepresents the relative importance of decision

criteria i and j. eAk ¼ eak ij h i (7) Step 4: Calculating the fuzzy weights

On the basis of the Lambda–Max method [64], we cal-culate the fuzzy weights after step 3. The Lambda–Max method is introduced as follows.

(a) When α = 1, we obtain eAk_m¼ aijm



nn, and when

α = 0, we obtain the lower bound eAkl ¼ aijl



nn

and the upper bound eAk_u¼ aiju



nn. The weight

vector in the AHP process can be derived as Wkm¼

Wk im
 , Wk l ¼ W k il
 , and Wk u¼ W k iu
 , i = 1, 2….. n. (b) We compute the two constants Mkl and M

k

u to

minimize the fuzziness of the weight by using Equations (8) and (9). We define the lower and upper bounds of the weight separately as Equations (10) and (11) and acquire the fuzzy weight matrix

for decision maker k, Weki ¼ W*ilk; Wkim; W*iuk

  , i = 1, 2….. n. Mkl ¼ min Wk im Wkil 1≤i≤n j g  (8) Mk_u¼ max W k im Wk iu 1≤i≤n j g  (9) Wk_l ¼ W
k_il; Wk_il ¼ Mk_lWk_il; i ¼ 1; 2; …; n (10) Wk_u ¼ W
k_iu; Wk_iu ¼ Mk uW k iu; i ¼ 1; 2; …; n (11)

Step 5: Integrating the fuzzy weights of decision makers

We apply the geometric average to obtain the aggregate of the fuzzy weights of decision makers per Table III and Equation (12) [66], in which eWi is the aggregated fuzzy

weight of criterion i and eWk_iis the fuzzy weight of criterion i.

e Wi¼ ∏ K k¼1 e Wki  1=K ; ∀k ¼ 1; 2; …; K (12)

Step 6: Ranking of criteria

A proximity coefficient was defined to obtain the ranking order of the decision elements. The proximity coefficient is defined as Equation (13), in which CCiis the weight for

criterion i. Additionally, d Wei; 0

and dþ Wei; 0

are the distance measurements between two fuzzy numbers that could be calculated per Equations (14) and (15).

CCi¼ d Wei; 0 dþ Wei; 1 þ d _e Wi; 0 ; i¼ 1; 2; …; n; 0≤CCi≤1 (13) Table I. Random index (RI).

n 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

RI 0.00 0.00 0.58 0.90 1.12 1.24 1.32 1.41 1.45 1.49 1.51 1.48 1.56 1.57 1.59

Table II. Triangular fuzzy numbers.

Linguistic variables Positive triangular fuzzy numbers Positive reciprocal triangular fuzzy numbers Extremely strong (9, 9, 9) (1/9, 1/9, 1/9) Intermediate (7, 8, 9) (1/9, 1/8, 1/7) Very strong (6, 7, 8) (1/8, 1/7, 1/6) Intermediate (5, 6, 7) (1/7, 1/6, 1/5) Strong (4, 5, 6) (1/6, 1/5, 1/4) Intermediate (3, 4, 5) (1/5, 1/4, 1/3) Moderately strong (2, 3, 4) (1/4, 1/3, 1/2) Intermediate (1, 2, 3) (1/3, 1/2, 1) Equally strong (1, 1, 1) (1, 1, 1)

Table III. Linguistic scales.

Linguistic variables Corresponding triangular fuzzy number

Very poor (1, 1, 1) Poor (1, 3, 5) Fair (3, 5, 7) Good (5, 7, 9) Very good (7, 9, 9)

d Wei; 0 ¼ ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1 3 W  il 0 2 þ Wim 0 2 þ Wiu 0 2 h i r (14) dþ Wei; 1 ¼ ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1 3 W  il 1 2 þ Wim 1 2 þ Wiu 1 2 h i r (15)

Step 7: Evaluating the biometric technologies With the results of a series of simple rankings, the weights of all criteria in each object of the hierarchy relative to the entire level directly above were obtained. These in turn were all ranked and were carried from the upper layer to the lower layer. After the aforementioned analytic process, the weight of each assessment criterion was determined for integrated assessment of the biomet-ric technologies.

5. RESEARCH ANALYSIS

The study applied the assessment research model to evalu-ate the top six emergent biometric technologies discussed in the IBG market report [13] and considered the potential biometric technologies under different objects by using BNP analysis based on the weights of each criterion obtained per FAHP. Furthermore, according to the data analysis, we discussed management implications. 5.1. Data analysis

This study queried experts in the biometricsfield who are familiar with biometric technology market conditions to assess the top six major biometric technologies to accom-plish the research purpose. We used step 1 in the previous section to verify the consistency of the 18 expert question-naires. There were 15 valid questionnaires whose values of

CI and CR were less than 0.1. We employed the data to ob-tain thefinal criteria weights of the assessment structure, per steps 3–6. The analysis results are presented in Table IV.

Technology assessment (0.407) is the most-emphasized object when evaluating biometric technologies, with metric competence (0.351) and the key elements of bio-metrics (0.242) ranking second and third, respectively. Nevertheless, biometric competence and the key elements of biometrics are greater than 0.5, but the technology assessment is not. This indicates that when evaluating bio-metric technologies, evaluators should still take the details of the target technology into account.

On the basis of the weights, the evaluation of biometric technologies relied on linguistic variables in Table III to express subjective judgments of the experts to reflect natu-ral human considerations. The experts were asked to iden-tify biometric technologies corresponding to each criterion; their opinions were later aggregated through the geometric average technique suggested by Buckley [61]. The fuzzy assessments of biometric technologies based on the evalu-ation criteria are presented in Table V. Next, the procedure of defuzzification locates the BNP value [66]. Therefore, it is used in this study. The BNP value of the fuzzy number Rican be found by Equation (16). LRi, MRi, and URiare

the lower, middle, and upper synthetic performance values of alternative i, respectively, and the calculations of each are as illustrated.

BNPi¼½ðURiLRiÞþ MRð iLRiÞ

.

3þ LRi; ∀i (16)

The BNP values of the six biometric technologies on each criterion are shown in Table VI. The ranking of the biometric technologies then proceeds on the basis of the value of the derived BNP for each of the biometric technologies.

By applying the simple additive weighting method to calculate thefinal score of each biometric technology, we calculated the BNP values multiplied by the weights in Table VI to produce the final evaluation of the six Table IV. Weights used in the evaluation of biometric technologies.

Object Weight Criteria Weight within object Aggregated weight Rank

Technology assessment 0.407 Technology merit 0.188 0.077 7 Business effect 0.367 0.149 1 Technology development potential 0.239 0.097 3

Risk 0.206 0.084 6

Biometric competence 0.351 Accuracy 0.241 0.085 5

Scale 0.158 0.055 9

Security 0.329 0.115 2

Privacy 0.272 0.095 4

Key elements of biometric 0.242 Enrollment 0.191 0.046 11 Biometric reference 0.174 0.042 13 Comparison and comparison errors 0.261 0.063 8

Networking 0.178 0.043 12

biometric technologies. For example, the score of face recognition is as follows: 2:28  0:077 þ 5:58  0:149 þ 3:71  0:097 þ 3:86  0:084 þ 2:97  0:085 þ 4:25  0:055 þ 5:60  0:115 þ 3:47  0:095 þ 3:21  0:046 þ 2:41  0:042 þ 2:00  0:063 þ 3:85  0:043 þ 2:76  0:047 ¼ 3:831

As Table VII indicates, among all six biometric technol-ogies, fingerprint recognition is the most preferred, followed by iris recognition and face recognition. 5.2. Management implications

Thefirst FAHP result indicates that within the technology assessment object, “business effect” (0.367) is the most critical criterion in biometric technology evaluations. According to the IBG report in 2009 [13], the major expec-tation of biometrics popularization and development is the progress of commercialization. As thefirst criterion over-all,“business effect” also ranks first within the technology assessment to further stress the necessity of the realization and promotion of biometrics.

The results identified “Security” (0.329) as the first priority criterion of biometric competence. The demand for reliable authentication techniques increased in the wake of heightened concerns about security; therefore, biometric technology is regarded as an effective approach for enhancing network security [24]. In some instances, biometrics could be integrated with passwords or tokens to strengthen the security offered by an authentication system. Thus, we can use biometrics to improve user convenience while enhancing security. This implies that security is beneficial in facilitating the popularization of biometric technologies.

“Comparison and comparison errors” (0.261) is the pre-dominant ingredient within the key elements of biometrics used to evaluate biometric technology. It is necessary to evaluate the setting of the threshold in identification systems for better matching because both failure to acquire and failure to enroll in the comparison process mean that the system can distinguish and extract the qualified charac-teristics of the user’s biometric. Failure to acquire and/or failure to enroll indicate that this person’s detected biomet-ric features may not have sufficient quality to verify and recognize. Alternatively, a convenience-focused applica-tion could adjust software or the mechanism to provide little or no denial of legitimate matches to allow a certain acceptable degree of impostors. Hence, customers would be more likely to accept biometrics [67].

Overall,“business effect” accounts for a 14.9% aggre-gated weight to further accentuate the challenge of the commercialization of biometric technologies, which is also the most critical part of evaluating the feasibility of bio-metrics. Additionally,“security” accounts for 11.5% and

T able V . Fuzzy evaluation of six biometric technologies on criteria. Criteria Face recognition Fingerprint recognition Iris recognition Speaker recognition Vascular pattern recognition Palm print recognition Technology merit (1.67, 2.35, 2.84) (3.02, 4.31, 5.69) (1.51, 2.26, 2.91) (4.12, 5.88, 7.75) (2.00, 2.90, 3.65) (2.36, 3.05, 3.94) Business effect (3.75, 5.73, 7.27) (2.98, 4.35, 5.42) (2.60, 3.69, 4.56) (2.53, 3.35, 4.03) (1.99, 2.74, 3.70) (2.98, 3.83, 4.87) Technology development potential (2.65, 3.62, 4.85) (2.37, 3.41, 4.24) (1.56, 2.22, 2.76) (4.66, 5.83, 7.67) (2.84, 4.30, 5.74) (4.34, 5.90, 7.09) Risk (2.91, 3.75, 4.91) (3.98, 5.38, 6.67) (3.54, 5.25, 6.34) (2.47, 3.14, 3.87) (1.50, 2.20, 2.68) (3.39, 4.66, 6.20) Accuracy (2.13, 3.06, 3.72) (2.54, 3.26, 4.16) (3.97, 5.06, 6.33) (1.72, 2.55, 3.15) (3.05, 4.36, 5.37) (1.76, 2.22, 2.69) Scale (3.02, 4.39, 5.35) (3.23, 4.57, 5.70) (4.43, 5.73, 6.95) (1.90, 2.85, 3.63) (1.51, 2.19, 2.74) (1.58, 2.26, 2.73) Security (4.19, 5.65, 6.96) (4.15, 5.86, 7.27) (2.13, 3.28, 4.06) (2.72, 3.70, 4.53) (2.67, 3.75, 4.79) (2.30, 3.33, 4.17) Privacy (2.33, 3.51, 4.57) (4.50, 5.77, 7.45) (4.02, 5.57, 7.12) (2.26, 3.10, 4.12) (2.47, 3.67, 4.47) (2.58, 3.87, 4.69) Enrollment (2.32, 3.31, 4.02) (3.55, 5.45, 7.30) (3.66, 5.08, 6.39) (1.59, 2.30, 2.87) (2.27, 2.86, 3.64) (4.74, 5.98, 7.25) Biometric reference (1.65, 2.43, 3.16) (2.77, 3.56, 4.31) (1.98, 3.07, 4.01) (2.01, 2.67, 3.21) (2.65, 3.49, 4.49) (3.75, 5.47, 7.31) Comparison and comparison errors (1.37, 1.99, 2.65) (2.79, 3.62, 4.42) (2.49, 3.42, 4.19) (1.93, 2.90, 3.56) (3.90, 5.67, 7.09) (1.35, 2.01, 2.45) Networking (3.00, 3.75, 4.81) (3.51, 4.86, 6.38) (3.21, 4.35, 5.85) (1.63, 2.09, 2.59) (2.70, 3.82, 4.59) (2.62, 3.55, 4.55) Personal biometric criteria (1.95, 2.80, 3.53) (1.83, 2.31, 3.05) (3.13, 4.56, 6.05) (1.60, 2.00, 2.47) (4.35, 5.76, 6.93) (2.66, 3.82, 4.64) Note: The fuzzy value of the six biometric technologies on each criterio n is prese nted in the brackets.

ranks second; it reflects that how to guarantee security will be a key task in biometrics. The development of biometric technologies should therefore focus on capabilities in the security area as well as on customer expectations for biometrics. There is a special criterion in biometrics, “privacy,” which accounts for 9.5% and ranks as fourth, playing a very important role in the biometrics growth path. Finally, according to the data analysis, we performed evaluations to determine which criteria of biometrics could facilitate its future development. These analyses could indicate chances to stimulate the growth of biometric technologies.

After completion of the biometrics evaluation and selec-tion model using BNP analysis, the six biometric technol-ogies were evaluated to determine those with the most potential and therefore most recommendable. The perfor-mance of each biometric technology is pairwise compared by our experts. In Table VII, fingerprint recognition (4.453) is the most likely biometric technology among all six, followed by iris recognition (3.973), face recognition (3.831), palm print recognition (3.782), vascular pattern recognition (3.717), and speaker recognition (3.294).

Furthermore, scores for each biometric are gathered by cell and by criteria in Table VI. These scores illustrate the achievement distribution of a specific criterion through biometrics. Some significant explanations can be inferred on the basis of the results of the BNP analysis in Table VI. Fingerprint recognition performs best overall among the

six biometric technologies because of its potential to meet the criteria in the objects, technology assessment and bio-metric competence. Iris recognition and face recognition also separately perform well on certain criteria according to Table VI, which is why they could score high in the FAHP. However, only the ranking offingerprint recogni-tion is the same as in the IBG report.

In reality, fingerprint recognition is also the top one, accounting for 45.9% of the non automated fingerprint identification system (non-AFIS) biometrics market, followed by face recognition at 18.5% and iris recognition at 8.3% [13]. Face recognition may be less competitive in the future because, in the present market, face recognition is the second most popular biometric technology [13]. We can see that it originally receives higher scores on two criteria, business effects and security, but is weaker on the others. Therefore, it is obvious that biometric technologies evaluated under different viewpoints will have different results. Because each biometric technology has its own supporting principles and mechanisms, it is hard to tell which biometric technology is superior on each criterion within this object. Hence, the diffe-rence in the scores is also close. This indicates that when discussing the advantages of utilizing biometric competence, iris recognition and fingerprint recognition play well com-pared with the others because of their high scores on criteria in this object, as presented in Table VI. This result shows that iris recognition development may have a chance to catch up from behind with its excellent biometric competences. Not surprisingly, iris recognition occupied third place in market revenue overall in 2009 [13]. As long as one biometric technology can achieve its biometric competences, it should create chances to enlarge market share and penetration

6. CONCLUSION

Today, biometrics has been vigorously promoted around the world as a means to strengthen network security and privacy [3] as well as to facilitate a new industry. Although biometrics has been applied in specific areas for decades, Table VI. Best non-fuzzy performance values of six biometric technologies on criteria.

Criteria Face recognition Fingerprint recognition Iris recognition Speaker recognition Vascular pattern recognition Palm print recognition Technology merit 2.28 4.34 2.23 5.92 2.85 3.12 Business effect 5.58 4.25 3.62 3.30 2.81 3.89 Technology development potential 3.71 3.34 2.18 6.06 4.29 5.78

Risk 3.86 5.34 5.04 3.16 2.13 4.75 Accuracy 2.97 3.32 5.12 2.48 4.26 2.22 Scale 4.25 4.50 5.70 2.79 2.15 2.19 Security 5.60 5.76 3.16 3.65 3.74 3.27 Privacy 3.47 5.91 5.57 3.16 3.53 3.71 Enrollment 3.21 5.44 5.04 2.25 2.92 5.99 Biometric reference 2.41 3.55 3.00 2.63 3.54 5.51 Comparison and comparison errors 2.00 3.61 3.37 2.80 5.55 1.94 Networking 3.85 4.92 4.47 2.11 3.70 3.57 Personal biometric criteria 2.76 2.40 4.58 2.02 5.68 3.71

Table VII. Ranking of biometric technologies.

Biometric technologies Score Rank

Face recognition 3.831 3 Fingerprint recognition 4.453 1 Iris recognition 3.973 2 Speaker recognition 3.294 6 Vascular pattern recognition 3.717 5 Palm print recognition 3.782 4

biometrics has gradually proliferated in customer and embedded electronic products to enhance security and pri-vacy [7]. To meet various technology assessment aspects, biometric technologies should be carefully assessed with regard to distinct features.

This study focuses on clarifying how differing evalua-tion objects determine relevant biometric technologies. We employ FAHP that clearly ranks the criteria of the objects built on the basis of a literature review. Using the analysis results, we assess biometric technologies with the perspectives in previous studies about factors affecting biometric evaluation and selection and conduct a system-atic ranking of the factors. We also organize and interpret these outcomes to form suggestions. The FAHP research method employed by this study is capable of revealing the relevance of the criteria and aids in the comprehension of them. The deployment of FAHP also allows relevant authorities to understand the importance of considering objects to determine development strategies. The derived re-sults disclose a synthetic conclusion for different dimensions. In sum, this research suggests that biometrics should enhance and support the ranked criteria to increase competitiveness. With the aforementioned, we have drawn several conclusions. First,fingerprint recognition received the highest score, followed by iris and face recognition. In the technology as-sessment and biometric competence objects, fingerprint recognition was the most recommendable biometric technology, sofingerprint recognition was determined to be the first priority. As shown in Table VI, fingerprint recognition did not stand out in each object, but on aver-age, it performed the best overall. Therefore, fingerprint recognition ranked first and fully shows its leadership in the biometric technology competition. As a result, finger-print recognition is regarded as the most complete biomet-ric technology, which has certain advantages in the market. This result also indicates thatfingerprint recognition still has room to improve and keep up on the criteria on which it was not the best. Producers should think more about strategies and measures to solidifyfingerprint recognition technology in the near future.

Furthermore, each cell in Table VI identifies scores for each alternative by criterion. These scores represent the performance distribution of a specific criterion across the biometric technologies. Several important explanations can be made regarding the results in Table VI. Iris recogni-tion actually performed the best in the biometric compe-tence object, and vascular pattern recognition especially met the requirements of the key elements of biometric objects. In other words, this indicates that relevant people could review each of the six biometrics from each viewpoint and obtain different explanations for specific purposes or certain applications. With this deduction, for example after the 9/11 terrorism attack, iris recognition be-came the primary recognition technology because it is the most reliable type of biometric and has advantages in the biometric competence object. Moreover, iris recognition always plays the second level in multi-biometric systems [28]. The future of the iris recognition system is better in

fields that demand rapid identification of individuals in a dynamic environment [27]. However, some considerations may be argued because of its low performance on some criteria, as shown in Table VI. As iris recognition improves on these criteria, it could increase its penetration. Finally, we found that the priority of face recognition was ranked third. In the present market, face recognition is the sec-ond-most-popular biometric technology [13]. Originally, face recognition had higher scores on the business effect, security, and scale criteria but was weaker on the other criteria. Therefore, when only evaluating and selecting based on the technology assessment object, face recognition is preferable. This leads to the obvious conclusion that biomet-ric technologies evaluated under different scenarios will have different results. Relevant persons could use these results as a lens to speculate how they could develop biometrics for commercialization. As long as one biometric technology can improve its advantage on criteria in this model, it creates the chance to enlarge the market share and penetration.

In conclusion, management researchers are faced with the issue of assessing advanced technology to predict which will be utilized. This study applies FAHP and BNP analysis in evaluating biometric technologies to sim-ulate how different evaluation objects affect biometric technology selection. With the weights of evaluation objects, we found that management aspects alone cannot determine the evaluation viewpoints. For example, the technology assessment object only accounted for 0.407. This enables researchers to recognize that technology assessment should focus more on the specifics of target technologies. Doing so could help them more comprehen-sively evaluate and select biometrics for network security.

REFERENCES

1. Gupta P, Carew S. Apple buys mobile security firm AuthenTec for $356 million. Retrieved from http://www. reuters.com/article/2012/07/27/us-authentec-acquisi-tion-apple-idUSBRE86Q0KD20120727, Jul 27, 2012. 2. Apple Press Info. Retrieved from http://www.apple. com/pr/library/2013/09/10Apple-Announces-iPhone-5s-The-Most-Forward-Thinking-Smartphone-in-the-World. html, Sep 10, 2013.

3. Yu FR, Tang H, Leung CMV, Liu J, Lung CH. Biometric-based user authentication in mobile ad hoc networks. Security and Communication Networks 2008; 1(1):5–16. 4. Adeoye SO. A survey of emerging biometric technol-ogies. International Journal of Computer Applications 2010; 9(10):1–5.

5. Jain AK, Nandakumar K, Nagar A. Biometric template security. EURASIP Journal on Advances in Signal Processing 2008; Article ID 579416 2008: 17. 6. Delac K, Grgic M. A survey of biometric recognition

methods. 46th International Symposium Electronics in Marine, ELMAR-2004, Zadar, Croatia, 16–18 June 2004.

7. Bhattacharyya D, Ranjan R, Alisherov AF, Choi MK. Biometric authentication: a review. International Jour-nal of u- and e- Service Science and Technology 2009; 2(3):13–28.

8. Jain AK, Ross A. Multibiometric systems, appeared in communication of the ACM. Special Issue on Multi-modal Interfaces 2004; 47(1):34–40.

9. Riley RA, Kleist VF. The biometric technologies business case: a systematic approach. Information Management & Computer Security 2005; 13(2): 89–105.

10. Tran TA, Daim T. A taxonomic review of methods and tools applied in technology assessment. Technological Forecasting and Social Change 2008; 75(9):1396–1405. 11. Shen YC, Chang SH, Lin GTR, Yu HC. A hybrid se-lection model for emerging technology. Technological Forecasting and Social Change 2010; 77:151–166. 12. Fleischer T, Decker M, Fiedeler U. Assessing emerging

technologies—methodological challenges and the case of nanotechnologies. Technological Forecasting and Social Change 2005; 72(9):1112–1121.

13. International Biometric Group. Biometrics Market and Industry Report 2009–2014. International Biometric Group: New York, 2009.

14. Saaty TL. The Analytic Hierarchy Process. McGraw-Hill: New York, 1980.

15. Biometrics Identity Management Agency. Biometrics Glossary version 4.0. Software Engineering Center CECOM Life Cycle Management Command, 2010. Retrieved from http://www.biometrics.dod.mil/Files/ Documents/Standards/BioGlossary.pdf [accessed 20 April 2013].

16. Jain AK, Ross A, Prabhakar S. An introduction to bio-metric recognition. IEEE Transactions on Circuits and Systems for Video Technology, Special Issue on Image- and Video-Based Biometrics 2004; 14(1):4–20. 17. Heyer R. Biometrics Technology Review 2008. Defence Science and Technology Organisation: Edinburgh, South Australia, publication DSTO-GD-0538, 2008. 18. NBSP. Biometric Technology Application Manual

Volume One, Biometric Basics Compiled and Published by: National Biometric Security Project 2008. Retrieved from http://www.planetbiometrics. com/creo_files/upload/article-files/btamvol1update.pdf [accessed 25 April 2013].

19. Dabbah MA, Woo WL, Dlay SS. Secure authentica-tion for face recogniauthentica-tion. In Proc. of IEEE Symposium on Computational Intelligence in Image and Signal Processing, Honolulu, HI, Apr. 2007; 121–126. 20. National Science and Technology Council.

Biomet-rics“Foundation Documents”. 2006. Retrieved from www.biometrics.gov/documents/biofoundationdocs.pdf [accessed 20 March 2013].

21. Goudelis G, Tefas A, Pitas I. Emerging biometric modalities: a survey. Journal on Multimodal User Interfaces 2008; 2(3–4):217–235.

22. Hong JH, Yun EK, Cho SB. A review of performance evaluation for biometrics systems. International Jour-nal of Image & Graphics 2005; 5(3):501–536. 23. Vielhauer C. Biometric modalities. Different traits for

authenticating subjects: chapter 3 in biometric user authentication for IT security. Advances in Information Security 2006; 18(I):33–75.

24. Jain AK, Ross A, Pankanti S. Biometric: a tool for information security. IEEE Transactions on Informa-tion Forensics and Security 2006; 1(2):125–144. 25. Xi K, Ahmad T, Han F, Hu J. Afingerprint based

bio-cryptographic security protocol designed for client/server authentication in mobile computing environment. Scurity and Communication Networks 2011; 4(5):487–499. 26. Ross A, Dass S, Jain AK. A deformable model for

fingerprint matching. Journal of Pattern Recognition 2005; 38(1):95–103.

27. Ross A. Iris recognition: the path forward. IEEE Com-puter 2010; 43(2):30–35.

28. Ganorkar RS, Ghatol AA. Iris recognition: an emerg-ing biometric technology. in Proc. of the 6th WSEAS International Conference on Signal Processing, Robotics and Automation, Greece, 2007; 91–96. 29. Kar B, Kartik B, Dutta PK. Speech and face biometric

for person authentication. In Proc. of IEEE Interna-tional Conference on Industrial Technology, India, 2006; 391–396.

30. Bhattacharyya D, Das P, Kim TH, Bandyopadhyay KS. Vascular pattern analysis towards pervasive palm vein authentication. Journal of Universal Computer Science 2009; 15(5):1081–1089.

31. Kong A, Zhang D, Kamel M. A survey of palm print rec-ognition. Pattern Recognition 2009; 42(7):1408–1418. 32. Hsu YG, Tzeng GH, Shyu JZ. Fuzzy multiple criteria

se-lection of government-sponsored frontier technology R&D projects. R&D Management 2003; 33(5):539–551. 33. Lee YG, Song YI. Selecting the key research areas in nano-technologyfield using technology cluster analy-sis: a case study based on national R&D programs in South Korea. Technovation 2007; 27(1–2):57–64. 34. Huang CC, Chu PY, Chiang YH. A fuzzy AHP

appli-cation in government-sponsored R&D project selec-tion. Omega-International Journal of Management Science 2008; 36(6):1038–1052.

35. Coldrick S, Longhurst P, Ivey P, Hannis J. A R&D options selection model for investment decisions. Technovation 2005; 25(3):185–193.

36. Shehabuddeen N, Probert D, Phaal R. From theory to practice: challenges in operationalising a technology se-lection framework. Technovation 2006; 26(3):324–407.

37. The Advanced Technology Program. How ATP Works. 2010. Retrieved from http://www.atp.nist.gov/ atp/overview.htm [accessed 25 March 2013]. 38. Yu OS, Hsu GYJ, Chen TY. Introduction to

Techno-logical Management: TechnoTechno-logical Forecast and Planning. Wu Nan Publishing Company: Taipei, 1998 (In Chinese).

39. Bolle RM, Connell JH, Pankanti S, Ratha NK, Senior AW. Guide to Biometrics. Springer: New York, 2004. 40. Kent S, Millett L. Who Goes There?: Authentication Technologies Through the Lens of Privacy. National Academies Press: Washington D.C., 2003.

41. Yang W, Hu J, Wang S. Afinger–vein based cancellable biocryptosystem. Network and System Security, Lecture Notes in Computer Science 2013; 7873:784–790. 42. Ahmad T, Hu J, Wang S. String-based cancelable

fingerprint templates. In Proc. of the 6th IEEE Confer-ence on Industrial Electronics and Applications (ICIEA), Beijing, 2011; 1028–1033.

43. Uludag U, Pankanti S, Prabhakar S, Jain AK. Biometric cryptosystems: issues and challenges. Proceedings of the IEEE, Special Issue on Enabling Security Technologies for Digital Rights Management 2004; 92(6):948–960. 44. Wang S, Hu J. Alignment-free cancellablefingerprint

template design: a densely infinite-to-one mapping (DITOM) approach. Pattern Recognition 2012; 45(12): 4129–4137.

45. Wei CC, Chien CF, Wang MJ. An AHP-based ap-proach to ERP system selection. International Journal of Production Economics 2005; 96:47–62.

46. Hu AH, Hsu CW, Kuo TC, Wu WC. Risk evaluation of green components to hazardous substance using FMEA and FAHP. Expert Systems with Applications 2009; 36(3):7142–7147.

47. Sloane EB, Liberatore MJ, Nydick RL, Luo W, Chung QB. Using the analytic hierarchy process as a clinical en-gineering tool to facilitate an iterative, multidisciplinary, microeconomic health technology assessment. Com-puters & Operations Research 2003; 30(10):1447–1465. 48. Vaidya OS, Kumar S. Analytic hierarchy process: An overview of applications. European Journal of Opera-tional Research 2006; 169(1):1–29.

49. Balestra G, Knaflitz M, Massa R, Sicuro M. AHP for the acquisition of biomedical instrumentation. In Conf. Proc. IEEE Eng. Med. Biol. Soc., Lyon, 2007; 3581–3584. 50. Gerdsri N, Kocaoglu DF. Applying the analytic

hierar-chy process (AHP) to build a strategic framework for technology road mapping. Mathematical and Com-puter Modelling 2007; 46(7/8):1071–1080.

51. Kim WK, Han SK, Oh KJ, Kim TY, Ahn HC, Song CW. The dual analytic hierarchy process to prioritize emerging technologies. Technological Forecasting and Social Change 2010; 77(4):566–577.

52. Kwong CK, Bai H. Determining the important weights for the customer requirement in QFD using a fuzzy AHP with an extent analysis approach. IIE Transac-tion 2003; 35:619–626.

53. Chan FTS, Kumar N. Global supplier development considering risk factors using fuzzy extended AHP-based approach. Omega 2007; 35(4):417–431. 54. Lee HI, Chen WC, Chang CJ. A fuzzy AHP and BSC

approach for evaluating performance of IT development in the manufacturing industry in Taiwan. Expert Systems with Applications 2008; 34(1):96–107.

55. Fu HP, Chao P, Chang TH, Chang YS. The impact of market freedom on the adoption of third-party elec-tronic marketplaces: a fuzzy AHP analysis. Industrial Marketing Management 2008; 37:698–712.

56. Gupta MM, Saridis GN, Gaines BR. Fuzzy Automata and Decision Processes. Elsevier North-Holland: New York, 1997.

57. Tsaur SH, Tzeng GH, Wang KC. Evaluating tourist risks from fuzzy perspectives. Annual of Tourism Research 1997; 24(4):796–812.

58. Zadeh LA. Fuzzy sets. Information and Control 1965; 8(3):338–353.

59. Deng H. Multicriteria analysis with fuzzy pairwise comparison. International Journal of Approximate Reasoning 1999; 21(3):215–231.

60. Laarhoven PJM, Pedrycz W. A fuzzy extension of Saaty’s priority theory. Fuzzy Sets and Systems 1983; 11(1–3):229–241.

61. Buckley JJ. Fuzzy hierarchical analysis. Fuzzy Sets and Systems 1985; 17(3):233–247.

62. Shen YC, Lin GTR, Li KP, Yuan BJC. An evaluation of exploiting renewable energy sources with concerns of policy and technology. Energy Policy 2010; 38:4604–4616.

63. Hsieh TY, Lu ST, Tzeng GH. Fuzzy MCDM approach for planning and design tenders selection in public office buildings. International Journal of Project Management 2004; 22(7):573–584.

64. Csutora R, Buckley JJ. Fuzzy hierarchical analysis: the Lambda–Max method. Fuzzy Sets and Systems 2001; 120(2):181–195.

65. Lee SK, Yoon YJ, Kim JW. A study on making a long-term improvement in the national energy efficiency and GHG control plans by the AHP approach. Energy Policy 2007; 35(5):2862–2868.

66. Opricovic S, Tzeng GH. Defuzzification within a multi criteria decision model. International Journal of Uncertainty Fuzziness and Knowledge-based Systems 2003; 11(5):635–652.

67. Grijpink JHAM. Trend report on biometrics: Some new insights, experiences and developments. Com-puter Law & Security Report 2008; 24:261–264.

APPENDIX

Questionnaire Information Table

No. Position Affiliation

1 Vice President Egis Technology Inc. (fingerprint authentication solution provider) 2 Sales Hitachi, Ltd. Taiwan Branch (_{finger–vein authentication solution provider)} 3 RD Director Face-Tek Technology Inc. (face recognition access control system provider) 4 Asia Region Manager Fingerprint Cards AB (FPC) Company

5 Algorithm R&D Senior Management iFLYTEK Co., Ltd. (speech and language information processing provider) 6 Assistant Professor Department of Electrical Engineering, Chang Gung University

7 Professor Department of Electrical Engineering, National Tsing Hua University 8 Smartphone FW Integration R&D Division Manager MediaTek Inc.

9 Professor Department of Computer Science, National Tsing Hua University 10 Honorary Professor Institute of Electro-Optical Engineering, National Chiao Tung University 11 RD Director D-Link Corporation

12 System Customer Service Assistant Management SYSTEM Corporation 13 Product Manager, Mobile Product Division Asus Corporation 14 Manager of Human Interface System

Development, RD Division

Acer Inc.

15 RD Hardware Manager HTC Corporation 16 Software Project Manager HTC Corporation

17 Associate Professor Department of Electrical Engineering, National Sun Yat-Sen University 18 Senior Engineer LG Electronics Inc., Iris Technology Division (iris authentication