Selection of RFID solution provider - Semantic Scholar

The current issue and full text archive of this journal is available at www.emeraldinsight.com/0368-492X.htm

K 42,3

448

Selection of RFID solution provider A fuzzy multi-criteria decision model with Monte Carlo simulation Kazim Sari International Logistics and Transportation Department, Beykent University, Istanbul, Turkey Abstract Purpose – The purpose of this paper is to provide a comprehensive framework to help managers of a business enterprise effectively evaluate candidate RFID solution providers and then select the most suitable one. Design/methodology/approach – The selection of an RFID solution provider is modeled as a new hybrid fuzzy multi-criteria decision making problem. The proposed decision model is based on integration of Monte Carlo simulation with fuzzy analytic hierarchy process (AHP) and fuzzy Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) methods. In addition, an illustrative case is used to exemplify the proposed approach. Findings – A quantitative methodology based on a structured framework, for the selection of the most appropriate RFID solution provider. Practical implications – This research study is a very useful source of information for managers of a business enterprise in making decisions about evaluation and selection of RFID solution providers or RFID system integrators. Originality/value – This study addresses the evaluation and selection of RFID solution providers for the managers of a business enterprise and proposes a new hybrid decision-making methodology for the problem. Keywords Radio frequency identification, Solution provider selection, Systems integrator selection, Fuzzy AHP, Fuzzy TOPSIS, Decision support systems, Monte Carlo simulation, Fuzzy logic Paper type Research paper

Kybernetes Vol. 42 No. 3, 2013 pp. 448-465 q Emerald Group Publishing Limited 0368-492X DOI 10.1108/03684921311323680

1. Introduction Radio frequency identification (RFID) is a generic term that is used to describe a technology that uses radio waves to identify and track people, animals, goods and products in transit. This technology is grouped under a broad category of automatic identification and data capture (AIDC) technologies. In practice, barcode technology is the most popular AIDC system because of its implementation simplicity and low cost. However, it has also some important limitations (Myerson, 2007, p. 1). For instance, it requires line of sight for operation and the data on a barcode is very limited in size and cannot be modified or added later. However, modern application processes such as patient care or supply chain integration, need more advanced capabilities which a barcode system cannot achieve. At this stage, RFID technology can take a role by adding value to these modern applications through its extended functionalities. In actual fact, RFID is not a newly developed technology; its roots can be traced back to Second World War. However, its

implementation in modern business processes was not possible until the early 2000s as this technology was not economical at those times. RFID technology has several advantages over barcode technology: multiple tag read capability, no line of sight requirement for the communication, large operating and communication range, and read and write capability of transponder memory are just a few examples of these advantages (Myerson, 2007, p. 43; Finkenzeller, 2010, p. 7). As a result, these attractive benefits create a growing interest for this technology in various industries. Namely, it is known that many companies from pharmaceutical industry, health care, logistics, and retail plan to investigate or implement this technology in the near future (Angeles, 2005; Roh et al., 2009). Nevertheless, in spite of its prominent advantages, ensuring a successful deployment of RFID technology is not an easy task. Namely, there are some barriers that make it difficult to implement this technology. For instance, high capital cost and a lack of technological maturity are two crucial factors that create reluctance for a business organization to use this technology (Attaran, 2011; Huber et al., 2007). Apart from these two factors, there are also a number of challenges that may be faced in the implementation process of an RFID project. In a recent paper, Cheung and Choi (2011) emphasized some of these implementation issues in designing an RFID-based anti-counterfeiting system. Similarly, Ngai et al. (2010) also elucidated the problems with RFID system implementation process along with a case study analysis of a textile dyeing and printing mill in China. In addition, in a trade magazine, Sullivan (2005) provided valuable information for us about the problem areas observed in the implementation processes of RFID projects deployed in different organizations such as Wal-Mart, Gilette, Kimberly-Clarck, and FedEx. Thus, as the above mentioned studies also indicate, managing RFID projects is a very difficult job for a business organization since an advanced level of engineering and technical capability along with in depth experience in RFID business is a must for the success. In addition, it is also the case that international standards and regulations have not been fully developed for this technology yet. That is, there are still many works on developing new standards and regulations in this area (“GS1 launches new RFID standard”, 2011, p. 16; “New ISO RFID standard will help trace products”, 2010, p. 18). Thus, keeping up the latest developments about this technology is very crucial for a business organization in order to be sure that a particular RFID system component is compatible with the new trends and standards in this area. Consequently, to manage this complicated situation, many business organizations today are required to get professional help from an outside company specialized in RFID business. These types of companies are called RFID solution providers or RFID systems integrators. Indeed, there is a variety of companies in this context that can provide full-scale services by managing all aspects of an RFID deployment project, from design to implementation and integration, and to the management of system upgrade and maintenance, for hiring organizations. Interested readers can see the online map at RFID Journal web site (www.rfidjournal.com) as a guide to see the deployments of RFID projects in different parts of the world along with their solution providers. As it is shown in this online map, there are a number of different solution providers or RFID system integrators worldwide with different sizes and expertise in different industries. In this research, we aim to develop a decision model to help managers of a business organization evaluating alternative RFID solution providers and then selecting the best suitable one for their organizations. As it is known, the role of a solution provider

Selection of RFID solution provider

449

K 42,3

450

in an RFID deployment project is very crucial. This is because, most of the time, deployment of RFID technology is not like one size fits all projects. That is, depending on the specific operational and environmental conditions of a hiring organization, the required hardware and software components as well as the integration units for these components might change substantially. Thus, for successful completion, an RFID project is needed to be customized very carefully by its solution provider. However, in spite of its importance level for the success of an RFID deployment project, we observe that there is very little research study carried out that devotes explicit attention to this issue. For instance, selection of the logistics partner or enterprise resource planning software is widely discussed by various researchers from different perspectives (Wei et al., 2005; Buyukozkan et al., 2008; Marasco, 2008; Cebeci, 2009; Sen et al., 2009). However, an equal attention has not been paid on the selection of RFID solution provider. In fact, this situation can partly be explained with the ongoing debates about the economical profitability of RFID technology in some industries (“RFID price tag too high for the corrugated box industry”, 2005; Beauchamp, 2008). Hence, in our opinion, in response to these reservations by the practitioners, the researchers in this area have focused more attention on exploring the potential benefits of this technology (Ustundag and Tanyas, 2009; Sari, 2010). As a result, to the best of our knowledge, only a few researchers (Cebeci and Kilinc, 2007; Radhika and Sattanathan, 2010; Wang et al., 2009, p. 519) concentrated on suggesting a model for RFID system or solution provider selection. Not surprisingly, a multi-criteria decision making (MCDM) approach is applied in all of these studies as the decision situation for this problem contains multiple and usually conflicting criteria. For instance, in their conference paper, Cebeci and Kilinc (2007) proposed a fuzzy analytic hierarchy process (AHP) model to select the best RFID system for glass industry. In a very similar environment, another conference paper by Radhika and Sattanathan (2010) proposed a fuzzy technique for order preference by the similarity to ideal solution (TOPSIS) model to determine the most appropriate RFID system. On the other hand, Wang et al. (2009) focused more attention on the selection of an RFID system supplier or solution provider for healthcare industry. In their model, a fuzzy TOPSIS based decision model is proposed for this purpose. Although these studies are very helpful for us to understand the decision situation more clearly, in their current state, they have some limitations to be used as a tool for the managers of a business enterprise. Among others, one important limitation is that they do lack to provide detailed evaluation criteria to be used for the selection of an appropriate RFID solution provider. More specifically, in these studies, a very general set of criteria such as quality, flexibility, and price are considered for the evaluation and selection purpose. As it is expected, these types of criteria are not able to reflect the specific business conditions at the project site and also the distinct application requirements for RFID technology. Thus, this research study aims to fill this lack in the literature by developing a comprehensive framework for the managers of a business enterprise. More specifically, the proposed decision model in this research aims to help managers of a business organization effectively evaluate candidate RFID solution providers and then select the best suitable one for their specific business conditions. For this purpose, the selection of RFID solution provider is modeled as a Monte Carlo simulation integrated MCDM problem, and then we present a novel approach to solve it.

In the evaluation procedure, two MCDM methods along with a Monte Carlo simulation analysis are used. These are fuzzy AHP and fuzzy extension of the TOPSIS. In the model, while fuzzy AHP is used to determine the relative weights of evaluation criteria, the fuzzy extension of TOPSIS method along with Monte Carlo simulation is used to select the best RFID solution provider. In fact, integration of fuzzy AHP and fuzzy TOPSIS is not a new approach in MCDM discipline. Namely, various research studies already indicated that integration of fuzzy AHP and fuzzy extension of TOPSIS approaches is very useful in providing viable solution alternatives (Torfi et al., 2010) for the evaluation and selection problems. However, the novelty of our proposed decision model lies in the integration of Monte Carlo simulation into the decision model. Indeed, the contribution of this integration is valuable for the managers of a business organization as it is now possible for them to see how the changes in relative weights of the evaluation criteria can influence the performance scores of the alternative RFID solution providers. The paper is then organized as follows. Section 2 describes the details of the proposed evaluation framework and the methods involved. An example case is given to demonstrate the potential of the methodology in Section 3. Finally, the last section contains some concluding remarks. 2. A performance evaluation framework for RFID solution provider selection The presented methodology in this paper is carried out into three phases. The first phase is related with extracting the evaluation and selection criteria for RFID solution providers. The second phase is concerned with weighting the RFID solution provider evaluation criteria using the fuzzy AHP method. Lastly, the third phase is the determining the best RFID solution provider using the fuzzy TOPSIS method along with a Monte Carlo simulation analysis. Figure 1 shows the detailed steps in these phases of the proposed methodology. 2.1 RFID solution provider evaluation criteria In fact, revealing the evaluation criteria to be used in this research study was a very difficult job for us as there is a very limited research study in this area. Therefore, to deal with this problem, we made an extensive review of the resources and the materials related with this topic. During this process, we observe that most of the works on this area is solely based on survey results of independent market research firms and expert opinions in RFID industry. These findings are summarized below. In July 2007 survey by ABI, seven RFID solution provider selection criteria are obtained from the end-user interviews. These are reported as price, vendor application specialization, vendor experience with RFID implementations, existing customer validation of vendor solutions, total solutions offerings, vendor service and support, return on investment, and total cost of ownership assessment tools (RFID Goes Mainstream, 2007). Indeed, a similar set of evaluation criteria for RFID solution providers is also reported by Manish and Moradpour (2005, p. 128). These are vital statistics, the management team, customer references, experience in RFID technology, experience in your industry, intellectual property rights, partnership and alliances, adherence to standards, and open architecture. On the other hand, another survey of ABI in fashion apparel and footwear industry reveals slightly a different set of criteria for RFID solution provider evaluation and


451

K 42,3

Form up the decision making team

Determine the alternative RFID solution providers

Determine the criteria to be used in the evaluation process

Structure the decision hierarchy

NO

Pre-processing phase

452

Approve the decision hierarchy?

Obtain the fuzzy criteria weights

Evaluate the alternative RFID solution providers

Make the Monte Carlo simulation analysis

Figure 1. Proposed methodology

Monte Carlo simulation integrated Fuzzy TOPSIS phase

Make pairwise comparisons for each dimension of the problem

Fuzzy AHP phase

YES

Select the best RFID solution provider

selection decisions. These are, as reported by Liard (2009), experience in full-scale deployments, ability to innovate, availability of viable products/solutions, dedication to the marketplace, financial stability, platform flexibility, and scalability. In addition, a research conducted by AMR in consumer products indicates that four criteria are very essential for the selection of an RFID solution provider in consumer products industry. These are reported as solution provider’s current RFID expertise,

its ability to deliver RFID services in the consumer products sector, the quantity and quality of its past RFID deployments in this industry, and its ability to support a global RFID implementation for an international consumer products manufacturer (Collins, 2004b). In parallel to this findings, another research shows that the number of RFID implementations completed before and whether or not a solution provider had implemented an RFID system at companies similar to the respondent’s own organization are the most influential factors for the positive perception of a solution provider (“Users tell RFID vendors: ‘show us the references’”, 2006). Thus, these two studies highlight the importance of having experience in a particular industry for those customers in that industry. In actual fact, this reality explains why solution providers are now getting more and more focus on individual industries in order to gain a competitive advantage over their competitors (Collins, 2004a). Lastly, the experts at RFID Journal (“Ten questions to ask your integrator”, 2005; Mapp, 2011) and at RFID Tribe (“Questions to ask your RFID systems integrator”, 2011) also suggest a set of evaluation criteria for RFID solution providers. As it is expected, their suggestions are more practitioners-oriented criteria. They are as follows: area of expertise, experience in AIDC technology, industry knowledge, ability to make business case analysis, having facitilies to test products, middleware platform used, ability to develop custom coding, vision for how to build on the system, intellectual property rights, references from key customers, and hardware used by the system integrator. In addition to these reports and the suggestions, we have also reviewed the related literature in similar areas (e.g. enterprise resource planning or third party logistics partner selection) for the purpose of understanding the factors that may be important for the selection of a suitable RFID solution provider (Wei et al., 2005; Buyukozkan et al., 2008; Marasco, 2008; Cebeci, 2009; Sen et al., 2009). To sum up, all of the above mentioned studies provided valuable resources for us to extract the evaluation and selection criteria for RFID solution providers. Namely, while determining the final set of evaluation criteria to be used for our research study, each criterion suggested by these works are carefully examined. Then, a comprehensive, yet manageable list of evaluation criteria is extracted. As a result, at the end of this process, two groups of evaluation criteria are determined for the proposed framework. Here, while the first group of criteria measures the performance of the RFID solution provider, the second group focuses on the properties of the RFID system provided by the solution provider. First group of criteria can be enumerated as follows: . Experience in RFID implementations (C 1 ). End-users want to see that the solution provider has managed a number of full-scale RFID projects. . Application specialization (C 2 ). End-users want to make sure that the solution provider has enough experience in their particular industry. . Customer references (C 3 ). The number and also the brand name of the references are very important for the end-users as they indicate the quality and timing of the work done before by the solution provider. . Technical/engineering capability (C4 ). End-users want to make sure that the solution provider has enough technical and engineering capability for system design and application. . Innovation capability (C5 ). End-users prefer to work with solution providers that have the capability to develop and implement new ideas and methods as the


453

K 42,3 .

454

.

requirements and expectations of different organizations can differ substantially during an RFID deployment project. Service and support capability (C6 ). After sales service and support activities are very crucial for the end-users as RFID systems need to be adopted for the changes in technology and standards. Financial stability (C 7 ). End-users need to have complete confidence when they assign budget to a major investment for an RFID project that the solution provider has enough financial strength to provide support.

Second group of criteria is as follows: . Total cost of ownership (C 8 ). Total cost of acquisition and operating cost as well as upgrade and replacement costs should be taken into consideration all together for an RFID system. . Platform flexibility (C 9 ). End-users prefer to implement an RFID system which does not require switching operating systems, platforms, and application software. . Scalability (C 10 ). End-users prefer to deploy an RFID system that is built around the correct set of standards to ensure that substantial expenditures will not be required for replacement or upgrades in the future. In addition, they also prefer that the system must be flexible enough to leverage the latest developments in this technology. Therefore, using these two groups of criteria, a decision model is created for the selection of RFID solution provider. Figure 2 shows the hierarchical structure of the decision model. The details of the model are explained in the following sections. 2.2 Fuzzy AHP method to obtain the weights of the evaluation criteria AHP is a popular method that is used to determine the relative importance of a list of evaluation criteria in a multi-criteria decision problem. It was first proposed by Saaty (1980) and then used by various researchers from different disciplines (Leung et al., 1998; Lee et al., 2008).

Figure 2. Hierarchical structure of the decision model

This method is based on three steps. In the first step, hierarchical structure of the decision model is constructed. Later, comparative judgments of the attributes and the criteria in each dimension of the decision hierarchy are formed by the expert opinions. Finally, synthesis of the priorities is performed. In the classical AHP method, the ratings and weights of criteria are measured in crisp numbers. However, under many conditions, crisp numbered data are inadequate to model the real life situations since human judgments and preferences of experts are often vague and imprecise. Therefore, to cope with this limitation of classical AHP, integration of fuzzy set theory (Zadeh, 1965) is proposed by Laarhoven and Pedrycz (1983). In the fuzzy AHP method, ratings and the weights of the criteria are expressed in linguistic terms and then set into fuzzy numbers. The most commonly used fuzzy numbers to capture the vagueness of linguistic assessments are triangular fuzzy numbers (TFNs) (Liang and Wang, 1994). In this research study, it is also proposed to use TFNs to represent the linguistic assessments of the experts in RFID industry. A TFN can be expressed as (l, m, u). The parameters l, m, and u indicate the smallest possible value, the most promising value, and the largest possible value, respectively. ~ and B~ be the two TFNs parameterized by the triplet (l1, m1, u1) and (l2, m2, u2), Let A respectively, then the operational laws of these two TFNs can be expressed as follows: ~ B~ ¼ ð l 1 þ l 2 ; m1 þ m2 ; u1 þ u2 Þ A%

ð1Þ

~ B~ ¼ ð l 1 l 2 ; m1 m2 ; u1 u2 Þ A^

ð2Þ

~ ¼ ð kl 1 ; km1 ; ku1 Þ k . 0; k^ðAÞ ~ 21 ¼ ðAÞ

1 1 1 ; ; u1 m1 l 1

k[R

ð3Þ

ð4Þ

~ and B~ can be calculated as In addition, the distance between two fuzzy numbers A follows (Chen, 2000): ~ BÞ ~ ¼ dðA;

rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1 ½ðl 1 2 l 2 Þ2 þ ðm 1 2 m2 Þ2 þ ðu1 2 u2 Þ2 3

ð5Þ

Therefore, given this information on fuzzy AHP method, in case there are n decision criteria/attributes (C1, C2, . . . , Cn) and K experts in an evaluation and selection problem, the procedure for determining the evaluation weights of each criterion in each dimension of a decision hierarchy can be explained as follows: Step 1: construct fuzzy pairwise comparison matrices. Through expert questionnaires, each decision maker is asked to assign a linguistic term to the pairwise comparisons among all criteria in the dimensions of a hierarchy system. The results of the comparisons ~k are constructed as fuzzy pairwise comparison matrices as shown in equation (6), where A is the fuzzy pairwise comparison matrix of kth decision maker:


455

K 42,3

456

2

1

6 k 6 a~ 21 6 k ~ A ¼6 6 .. 6 . 4 a~ kn1

a~ k12

...

1

...

.. .

..

a~ kn2

...

.

a~ k1n

3

2

1

7 6 a~ k2n 7 6 1=~ak12 7 6 6 .. 7 7 ¼ 6 .. 7 6 . . 5 4 1 1=~ak1n

a~ k12

...

1

...

.. .

..

1=~ak2n

...

.

a~ k1n

3

7 a~ k2n 7 7 .. 7 7; . 7 5 1

k ¼ 1; 2; . . . ; K ð6Þ

Step 2: compute the synthetic pairwise comparison matrix. After collecting the expert opinions, the next step is to combine them together. For this purpose, geometric mean technique is used as shown in equation (7), where a~ ij is the aggregated fuzzy comparison value of dimension i to criterion j, and K is the total number of experts: 1=K ð7Þ a~ ij ¼ a~ 1ij ^~a2ij ^· · ·^~aK ij Step 3: compute the fuzzy weight of each criterion. At this step, fuzzy weight of each criterion is obtained by using the geometric mean method suggested by Buckley (1985). The geometric mean of fuzzy comparison value of criterion i to each criterion (~ri ) can be found by equation (8): ð8Þ r~i ¼ ð~ai1 ^· · ·^~aij ^· · ·^~ain Þ1=n Then, the fuzzy weight of the ith criterion (w~ i ) indicated by a TFN, w~ i ¼ ðlwi ; mwi ; uwi Þ can be obtained by equation (9): w~ i ¼ r~i ^ð~r1 %· · ·%~ri %· · ·%~rn Þ21

ð9Þ

2.3 Monte Carlo simulation integrated fuzzy TOPSIS method As it is known, TOPSIS method was first proposed by Hwang and Yoon (1981). According to this technique, the best alternative would be the one that is nearest to the positive ideal solution and farthest from the negative ideal solution. At this point, while the positive ideal solution represents a solution that maximizes the benefit criteria and minimizes the cost criteria, the negative ideal solution maximizes the cost criteria and minimizes the benefit criteria (Chen, 2000). Thus, in selecting the most suitable alternative, the TOPSIS method considers simultaneously the distances to both positive and negative ideal solutions. The fuzzy version of the TOPSIS method, on the other hand, is the TOPSIS method that is extended to fuzzy environment to deal with the imprecise and vague information. In this research, a Monte Carlo simulation analysis is integrated into the fuzzy TOPSIS method in order to better analyze the expert opinions. The basic steps of our proposed Monte Carlo simulation integrated fuzzy TOPSIS method are described as follows. Suppose that there are m alternatives (A1, A2, . . . , Am) and n decision criteria/attributes (C1, C2, . . . , Cn). Step 1: determine the weightings of the evaluation criteria. In his research, we have employed fuzzy AHP to find the fuzzy preference weights. As it is shown in Section 2.2, the fuzzy weights of each evaluation criterion (w~ i ) is expressed by a TFN, w~ i ¼ ðlwi ; mwi ; uwi Þ. In our proposed model, we have converted these TFNs (e.g. w~ i ¼ ðlwi ; mwi ; uwi Þ) to random numbers (ti) that come from a triangular probability distribution function with parameters (lwi, mwi, uwi) where (lwi, uwi) is the range and mwi is the most likely value.

Later, these random numbers are used to conduct a Monte Carlo simulation analysis in order to better understand the impact of variability or uncertainty in the weights of evaluation criteria on the model results. Step 2: choose the appropriate linguistic variables for the alternatives with respect to criteria. The linguistic variables are described by TFNs. Step 3: construct the fuzzy decision matrix. To obtain the fuzzy decision matrix shown in equation (10), the expert ratings for each alternative is aggregated with respect to each criterion as shown in equation (11):

A1 A D~ ¼ 2 .. . Am

2 6 6 6 6 6 6 4

C1

C2

...

x~ 11

x~ 12

...

x~ 21

x~ 11

...

.. .

.. .

..

x~ m1

x~ m2

...

7 a~ k2n 7 7 .. 7 7 . 7 5 x~ mn

x~ ij ¼

Cn x~ 1n

.

K

3 i ¼ 1; 2; . . . ; m;

1 x~ 1ij %~x2ij %· · ·%~xkij

j ¼ 1; 2; . . . ; n

ð10Þ

ð11Þ

where x~ kij is the performance rating of alternative Ai with respect to criterion Cj evaluated by the kth expert, and x~ kij ¼ ðl kij ; mkij ; ukij Þ. Step 4: normalize the fuzzy decision matrix. To obtain the normalized fuzzy decision ~ matrix denoted by R: R~ ¼ ½~rij mxn ;

i ¼ 1; 2; . . . ; m;

j ¼ 1; 2; . . . ; n

ð12Þ

Equations (13)-(16) are employed. Here, B and C are the set of benefit criteria and cost criteria, respectively: ! l ij mij uij ; þ; þ ; j[B ð13Þ r~ij ¼ uþ uj uj j 2 2 2 lj lj lj ; j[C ð14Þ r~ij ¼ ; ; uij mij l ij uþ j ¼ max

i{uij ji¼1;2; ... ;n}

l2 j ¼ min

if j [ B

if j [ C

i{l ij ji¼1;2; ... ;n}

ð15Þ ð16Þ

Step 5: calculate the weighted normalized decision-matrix. The weighted normalized decision matrix denoted by V~ is calculated by the following equations: V~ ¼ ½~vij mxn ;

i ¼ 1; 2; . . . ; m; v~ ij ¼ r~ij ^t i

j ¼ 1; 2; . . . ; n

ð17Þ ð18Þ


457

K 42,3

458

The preference weights of criteria (ti) are the random numbers obtained in Step 1. In fact, for each run of the simulation model, there will be a different weighted normalized decision-matrix. Step 6: determine the fuzzy positive-ideal solution (Aþ ) and fuzzy negative-ideal solution (A2 ): ~þ ~þ A þ ¼ v~ þ ð19Þ where v~ þ n 1 ;v 2 ; ...;v j ¼ ð1; 1; 1Þ ~2 ~2 A 2 ¼ v~ 2 where v~ 2 ð20Þ 1 ;v 2 ; ...;v n j ¼ ð0; 0; 0Þ Step 7: calculate the distance of each alternative from Aþ and A2 . The vertex method 2 þ 2 shown in equation (5) is employed to calculate the distances (d þ as: i ; d i ) from A and A dþ i ¼

n X d v~ ij ; v~ þ j ;

i ¼ 1; 2; . . . ; m;

j ¼ 1; 2; . . . ; n

ð21Þ

i ¼ 1; 2; . . . ; m;

j ¼ 1; 2; . . . ; n

ð22Þ

j¼1

d2 i ¼

n X d v~ ij ; v~ 2 j ; j¼1

Step 8: calculate the closeness coefficient of each alternative. The closeness coefficient of alternative Ai is defined as CCi: d2 ð23Þ CC i ¼ þ i 2 ; i ¼ 1; 2; . . . ; m di þ di Step 9: perform a simulation output analysis and then rank the alternatives according to the closeness coefficient (CCi) values. 3. Application of the proposed framework to an illustrative case In this section, the proposed decision model is applied to an illustrative case. For this purpose, a manufacturing company (ABC Company) that intends to deploy an RFID system for its main distribution operations is considered. Currently, ABC Company suffers from very high rate of errors in the inventory records at the main distribution center. As it is known, a very high rate of inaccuracy in inventory records can result in very serious problems for the design and daily operations of a supply chain (Fleisch and Tellkamp, 2005; Sari, 2008). For this reason, top managers of ABC Company plan to deploy an RFID system in the distribution center to deal with these errors in the inventory information. They believe that adoption of RFID technology will create a substantial improvement for the distribution channel operations (Ustundag and Tanyas, 2009; Sari, 2008). However, the company has some difficulties at this stage. Namely, while there are four candidate RFID solution providers to manage this project, top managers of ABC Company is unsure about which one of these solution providers is the most appropriate for their specific business and operational conditions. Therefore, to help top managers of this company in this complex decision situation, our proposed framework is applied. 3.1 The fuzzy weights of evaluation criteria The weights of the criteria to be used in evaluation process are calculated by using the fuzzy AHP method presented in Section 2.2. For this purpose, we propose the following linguistic weighting set:


{EI; WMI; SMI; VSMI; AMI}; where: EI

equally important.

WMI weakly more important. SMI

strongly more important.

459

VSMI very strongly more important. AMI

absolutely more important.

Table I shows the numeric conversions of these linguistic variables as well as the related fuzzy inverse conversion scale. For instance, one may consider that the criterion i is “very strongly more important” as compared with the criterion j under a certain dimension. Then, this linguistic expression can be converted to a fuzzy value as a~ ij ¼ ð2; 5=2; 3Þ. In a recent research study, the same conversion scale is also used by Buyukozkan et al. (2008). A group of five experts in the area of RFID technology and implementation are used to construct individual pairwise comparison matrices for each dimension of the problem. For our case, since there are two main sets of criteria as vendor dimension and system dimension, three different pairwise comparison matrices are constructed by each expert (Figure 2). At this point, while the first matrix is used to express the relative importance of the vendor dimension over the system dimension, second and third matrices are constructed to understand the local importance level of each criterion within their respective dimensions. As an example, pairwise comparison matrices for the main evaluation criteria are formed as follows: Experts 1 and 5 indicate that the vendor dimension is very strongly important (VSMI) than the system dimension; the other experts, on the other hand, state that the vendor dimension is absolutely more important (AMI) than the system dimension. Then, by using the corresponding fuzzy numbers, the evaluation matrices relevant to the main objective can be constructed as shown in Table II. Later, the next step is to obtain the synthetic pairwise comparison matrices for each dimension of the problem. For this purpose, geometric means of the individual expert opinions are found to obtain the synthetic pairwise comparison matrices as indicated in equation (7). For our case, three synthetic pairwise comparison matrices are formed for the main evaluation criteria, the system dimension, and the vendor dimension. As an example, Table III shows the synthetic comparison matrix of the main evaluation criteria. Afterwards, based on the synthetic comparison matrices, relative importance of each evaluation criteria can be calculated by following the AHP method explained in Section 2.2. As an example, by using equations (8) and (9), importance weights of the main evaluation criteria are calculated as (0.611, 0.736, 0.879) for the vendor dimension and (0.223, 0.264, 0.321) for the system dimension. In a similar fashion, fuzzy global Linguistic scale for importance degrees

Triangular fuzzy scale

Triangular fuzzy reciprocal scale

Equally important (EI) Weakly more important (WMI) Strongly more important (SMI) Very strongly more important (VSMI) Absolutely more important (AMI)

(1/2, 1, 3/2) (1, 3/2, 2) (3/2, 2, 5/2) (2, 5/2, 3) (5/2, 3, 7/2)

(2/3, 1, 2) (1/2, 2/3, 1) (2/5, 1/2, 2/3) (1/3, 2/5, 1/2) (2/7, 1/3, 2/5)

Table I. Triangular fuzzy conversion table

K 42,3

Experts/dimensions Expert 1 Expert 2

460

Expert 3

Table II. Fuzzy evaluation matrices of the main evaluation criteria

Expert 4

Table III. Synthetic comparison matrix of the main evaluation criteria

Expert 5

Vendor dimension System dimension Vendor dimension System dimension Vendor dimension System dimension Vendor dimension System dimension Vendor dimension System dimension

Vendor dimension System dimension

Vendor dimension

System dimension

(1, 1, 1) (1/3, 2/5, 1/2) (1, 1, 1) (2/7, 1/3, 2/5) (1, 1, 1) (2/7, 1/3, 2/5) (1, 1, 1) (2/7, 1/3, 2/5) (1, 1, 1) (1/3, 2/5, 1/2)

(2, 5/2, 3) (1, 1, 1) (5/2, 3, 7/2) (1, 1, 1) (5/2, 3, 7/2) (1, 1, 1) (5/2, 3, 7/2) (1, 1, 1) (2, 5/2, 3) (1, 1, 1)

Vendor dimension

System dimension

(1.00, 1.00, 1.00) (0.30, 0.36, 0.44)

(2.29, 2.79, 3.29) (1.00, 1.00, 1.00)

weights of the ten evaluation criteria for RFID solution providers can also be calculated. The results of these calculations are shown in Figure 3. At the end of this process, in order to use in the Monte Carlo simulation analysis, the fuzzy global weights shown in Figure 3 are converted to random numbers that come from a triangular probability distribution with the respective parameters shown in Figure 3. 3.2 The selection of the most suitable RFID solution provider Now, given the evaluation criteria structure and the related importance weights obtained in the previous section, the next step is to evaluate the four candidate RFID solution providers for this project and then select the best one. For this purpose, the Monte Carlo simulation integrated fuzzy TOPSIS method proposed in Section 2.3 is applied. To perform the Monte Carlo simulation analysis, Oracle Crystal Ball for Enterprise Performance Management software is used. As it is known, it is a popular spreadsheet based risk analysis and simulation analysis tool. In the Monte Carlo

Vendor Dimension (0.611, 0.736, 0.879)

Figure 3. Global importance weights of the ten RFID solution provider evaluation criteria

Evaluationg the Performance of RFID Solution Providers System Dimension (0.223, 0.264, 0.321)

Experience in RFID implementation Application specialization Customer references Technical/engineering capability Innovation capability Service and support capability Financial stability

Global Weights (0.043, 0.102, 0.218) (0.044, 0.102, 0.230) (0.084, 0.176, 0.351) (0.043, 0.096, 0.223) (0.035, 0.078, 0.183) (0.055, 0.118, 0.273) (0.032, 0.064, 0.120)

Total cost of ownership Platform flexibility Scalability

Global Weights (0.083, 0.139, 0.233) (0.036, 0.063, 0.113) (0.036, 0.061, 0.115)

simulation analysis, 10,000 simulation runs are conducted in order to eliminate the impact of random variations. The steps in the Monte Carlo simulation integrated fuzzy TOPSIS method are explained as follows. First, the candidate RFID solution providers are rated by the same five experts according to each criterion with linguistic variables. These linguistic variables and related numeric conversions are presented in Table IV. The same fuzzy conversion table is also used by Wang and Chang (2007). Based on the individual ratings of the five experts for each alternative, aggregated fuzzy decision matrix is formed by using equation (11). The obtained aggregated fuzzy decision matrix is presented in Table V. After constructing the aggregated fuzzy decision matrix, the next step is to normalize the fuzzy decision matrix and then calculate the normalized weighted fuzzy decision matrix. In fact, this point is where the Monte Carlo simulation analysis is actually performed. At this stage, while calculating the normalized weighted fuzzy decision matrix, the random numbers generated in the simulation model are used as the weights of the evaluation criteria. As it is known, these random numbers are defined by using the fuzzy global weights of the evaluation criteria obtained from the fuzzy AHP method. Afterwards, for each run of the Monte Carlo simulation, the distances of each 2 candidate from fuzzy positive ideal solution (d þ i ) and fuzzy negative ideal solution (d i ) þ are calculated. Fuzzy positive ideal solution (A ) and fuzzy negative ideal solution (A 2 ) are calculated by using equations (19) and (20). Then, the relative closeness to the ideal solution (CCi) is calculated for each RFID solution provider. At the end of the 10,000 simulation runs, frequency distributions of CCi for each RFID solution provider are created to make a decision about the RFID solution providers.

Linguistic scale for evaluating performance of solution providers

C1 C2 C3 C4 C5 C6 C7 C8 C9 C10

(0, 1, 3) (1, 3, 5) (3 5, 7) (5, 7, 9) (7, 9, 10)

RFID solution provider A (5.80, 7.80, 9.40) (5.80, 7.80, 9.40) (5.40, 7.40, 9.20) (4.20, 6.20, 8.20) (3.80, 5.80, 7.80) (4.60, 6.60, 8.60) (7.00, 9.00, 10.00) (5.80, 7.80, 9.40) (0.60, 2.20, 4.20) (4.20, 6.20, 8.20)

RFID solution provider B (0.60, (6.20, (3.80, (4.60, (3.00, (4.20, (1.40, (1.40, (6.20, (3.40,

2.20, 8.20, 5.80, 6.60, 5.00, 6.20, 3.40, 3.40, 8.20, 5.40,

4.20) 9.60) 7.80) 8.60) 7.00) 8.20) 5.40) 5.40) 9.60) 7.40)

RFID solution provider C (4.60, (4.20, (1.80, (3.80, (1.80, (1.40, (3.40, (3.40, (3.00, (5.80,

461

Triangular fuzzy scale

Very low (VL) Low (L) Medium (M) High (H) Very high (VH)

Evaluation criteria


6.60, 6.20, 3.80, 5.80, 3.80, 3.40, 5.40, 5.40, 5.00, 7.80,

8.60) 8.20) 5.80) 7.80) 5.80) 5.40) 7.40) 7.40) 7.00) 9.40)

Table IV. Linguistic scale for evaluating performance of solution providers

RFID solution provider D (5.40, (6.60, (5.40, (3.40, (6.20, (5.40, (6.60, (0.60, (5.80, (3.00,

7.40, 8.60, 7.40, 5.40, 8.20, 7.40, 8.60, 2.20, 7.80, 5.00,

9.20) 9.80) 9.20) 7.40) 9.60) 9.20) 9.80) 4.20) 9.40) 7.00)

Table V. Aggregated fuzzy decision matrix

K 42,3

462

To this end, Figure 4 is created. This figure indicates the overlay chart for the frequency distributions of CCi for each candidate RFID solution provider. Now, based on Figure 4, we can rank the solution providers from best to the worst as Vendor D, Vendor A, Vendor B, and Vendor C. Indeed, while making this rank, we are sure and confident about our decision as Figure 4 shows not only the average performance scores, but also all situations for each solution provider. For example, Figure 4 shows that Vendor D is the best alternative on the average. However, in addition to this information, it is also shown in Figure 4 that there is a small probability that Vendor A is better than Vendor D. Thus, with this additional information, the decision makers can better analyze the problem situation and then make more precise decisions. To sum up, with our proposed decision model, while making their final decisions, the managers of a business organization do not need to depend only on the average performance scores. In fact, this is an advantage of our proposed decision model as it provides all related information about the evaluation and selection problem for the decision makers. 4. Conclusion This research aims to help managers of an organization effectively evaluate candidate RFID solution providers and then select the most suitable one for their specific business conditions. For this purpose, the selection of RFID solution provider is modeled as a MCDM problem, and then we present a novel approach to solve it. In the evaluation procedure, two MCDM methods along with a Monte Carlo simulation model are used. These are fuzzy AHP and fuzzy extension of the TOPSIS. In the model, while fuzzy AHP is used to determine the relative weights of evaluation criteria, the fuzzy extension of TOPSIS along with the Monte Carlo simulation analysis is used to select the best RFID solution provider. Here, Monte Carlo simulation analysis is integrated into the proposed decision model for the purpose of analyzing the sensitivity of the selected alternative to the relative weights of evaluation criteria. By this way, the managers of a business enterprise can better understand the vague and

Figure 4. Overlay chart for the closeness to ideal solution (CCi) for the solution providers

imprecise information provided by the experts in this area. In fact, this is an important methodological contribution of our research. In addition, another contribution of our research study lies on the fact that with this research an inclusive set of evaluation criteria for RFID solution providers is extracted for the decision makers in this area. This is because our review of literature indicated the fact that there is very limited research study on this area for the practitioners. Therefore, the evaluation criteria obtained by this research study can be very useful for the business organizations that intend to use this technology. Finally, an illustrative case is employed to exemplify the proposed framework for the managers of a business organization. This illustrative case analysis shows the feasibility and practicability of our proposed model for real life applications. In fact, we believe that after this research study, managers of a business organization can make better analysis in evaluating candidate RFID solution providers and then select the most suitable solution provider. As a future step this research study could be the comparison of the proposed approach to other MCDM methods, such as ANP, DEA or genetic algorithms. References Angeles, R. (2005), “RFID technologies: supply-chain applications and implementation issues”, Information Systems Management, Vol. 22 No. 1, pp. 51-65. Attaran, M. (2011), “The supply demand for RFID”, Industrial Engineer, December, pp. 26-31. Beauchamp, M. (2008), “Don’t rule out the bar code”, Dairy Industries International, Vol. 73 No. 5, pp. 33-35. Buckley, J.J. (1985), “Fuzzy hierarchical analysis”, Fuzzy Sets Systems, Vol. 17 No. 1, pp. 233-247. Buyukozkan, G., Feyzioglu, O. and Nebol, E. (2008), “Selection of the strategic alliance partner in logistics value chain”, International Journal of Production Economics, Vol. 113 No. 1, pp. 148-158. Cebeci, U. (2009), “Fuzzy AHP-based decision support system for selecting ERP systems in textile industry by using balanced scorecard”, Expert Systems with Applications, Vol. 36 No. 5, pp. 8900-8909. Cebeci, U. and Kilinc, S. (2007), “Selecting RFID systems for glass industry by using fuzzy AHP approach”, RFID Eurasia, September 5-6, pp. 1-4. Chen, C.T. (2000), “Extension of TOPSIS for group decision-making under fuzzy environment”, Fuzzy Sets and Systems, Vol. 114, pp. 1-9. Cheung, H.H. and Choi, S.H. (2011), “Implementation issues in RFID-based anti-counterfeiting systems”, Computers in Industry, Vol. 62 No. 7, pp. 708-718. Collins, J. (2004a), “RFID consultancies go vertical”, RFID Journal, available at: www.rfidjournal. com/article/view/1096 Collins, J. (2004b), “Study ranks RFID implementers”, RFID Journal, available at: www. rfidjournal.com/article/view/942 Finkenzeller, K. (2010), RFID Handbook, Wiley, Chichester. Fleisch, E. and Tellkamp, C. (2005), “Inventory inaccuracy and supply chain performance: a simulation study of a retail supply chain”, International Journal of Production Economics, Vol. 95 No. 3, pp. 373-385. “GS1 launches new RFID standard” (2001), Material Handling & Logistics, Vol. 66 No. 10, p. 16.


463

K 42,3

464

Huber, N., Michael, K. and McCathie, L. (2007), “Barriers to RFID adoption in the supply chain”, RFID Eurasia, 2007 1st Annual, pp. 1-6. Hwang, C.L. and Yoon, K. (1981), Multiple Attributes Decision Making Methods and Applications, Springer, Berlin. Laarhoven, P.J.M. and Pedrycz, W. (1983), “A fuzzy extension of Saaty’s priority theory”, Fuzzy Sets and Systems, Vol. 11 Nos 1-3, pp. 229-241. Lee, S.K., Mogi, G. and Kim, J.W. (2008), “The competitiveness of Korea as a developer of hydrogen energy technology: the AHP approach”, Energy Policy, Vol. 36 No. 4, pp. 1284-1291. Leung, P.S., Muraoka, J., Nakamoto, S.T. and Pooley, S. (1998), “Evaluating fisheries management options in Hawaii using analytic hierarchy process (AHP)”, Fisheries Research, Vol. 36 Nos 2/3, pp. 171-183. Liang, G.S. and Wang, M.J. (1994), “Personnel selection using fuzzy MCDM algorithm”, European Journal of Operational Research, Vol. 78, pp. 22-33. Liard, M. (2009), RFID Item-level Tagging in Fashion Apparel and Footwear, ABI Research, Oyster Bay, NY, 4Q. Manish, B. and Moradpour, S. (2005), RFID Field Guide: Deploying Radio Frequency Identification Systems, Prentice-Hall, Upper Saddle River, NJ. Mapp, W. (2011), “The value of partnership”, RFID Journal, available at: www.rfidjournal.com/ article/articleview/8297/1/82 Marasco, A. (2008), “Third party logistics: a literature review”, International Journal of Production Economics, Vol. 113 No. 1, pp. 127-147. Myerson, J.M. (2007), RFID in the Supply Chain: A Guide to Selection and Implementation, Auerbach Publications, New York, NY. “New ISO RFID standard will help trace products” (2010), Official Board Markets, Vol. 86 No. 9, p. 18. Ngai, E.W.T., To, C.K.M., Moon, K.K.L., Chan, L.K., Yeung, P.K.W. and Lee, M.C.M. (2010), “RFID systems implementation: a comprehensive framework and a case study”, International Journal of Production Research, Vol. 48 No. 9, pp. 2583-2612. “Questions to ask your RFID systems integrator” (2011), RFID Tribe, available at: www.rfidtribe. com/index.php?option¼com_content&task¼view&id¼481&Itemid¼102 (accessed March 15, 2011). Radhika, R. and Sattanathan, R. (2010), “Selection of best RFID system using fuzzy TOPSIS”, Proceedings of the International Conference on Information Science and Applications (ICISA 2010), Chennai, India. RFID Goes Mainstream (2007), “RFID goes mainstream – an allian technology viewpoint”, White Paper, available at: www.alientechnology.com/docs/WP_Alien_Viewpoint.pdf (accessed March 15, 2011). “RFID price tag too high for the corrugated box industry” (2005), Industry Week, available at: www.industryweek.com/ReadArticle.aspx?ArticleID¼10321 Roh, J.J., Kunnathur, A. and Tarafdar, M. (2009), “Classification of RFID adoption: an expected benefit approach”, Information & Management, Vol. 46 No. 6, pp. 357-363. Saaty, T.L. (1980), The Analytic Hierarchy Process, McGraw-Hill, New York, NY. Sari, K. (2008), “Inventory inaccuracy and performance of collaborative supply chain practices”, Industrial Management & Data Systems, Vol. 108 No. 4, pp. 495-509.

Sari, K. (2010), “Exploring the impacts of radio frequency identification (RFID) technology on supply chain performance”, European Journal of Operational Research, Vol. 207 No. 1, pp. 174-183. Sen, C.G., Baracli, H., Sen, S. and Basligil, H. (2009), “An integrated decision support system dealing with qualitative and quantitative objectives for enterprise software selection”, Expert Systems with Application, Vol. 36 No. 3, pp. 5272-5283. Sullivan, L. (2005), “RFID implementation challenges persist, all this time later”, InformationWeek, October 10, available at: www.informationweek.com/news/171203904 “Ten questions to ask your integrator” (2005), RFID Journal, available at: www.rfidjournal.com/ article/view/1331 Torfi, F., Farahani, R.Z. and Rezapour, S. (2010), “Fuzzy AHP to determine the relative weights of evaluation criteria and fuzzy TOPSIS to rank the alternatives”, Applied Soft Computing, Vol. 10 No. 2, pp. 520-528. “Users tell RFID vendors: ‘show us the references’” (2006), RFID Journal, available at: www. rfidjournal.com/article/view/6626 (accessed March 17, 2011). Ustundag, A. and Tanyas, M. (2009), “The impacts of radio frequency identification (RFID) technology on supply chain costs”, Transportation Research Part E, Vol. 45 No. 1, pp. 29-38. Wang, T.-C. and Chang, T.-H. (2007), “Application of TOPSIS in evaluating initial training aircraft under a fuzzy environment”, Expert Systems with Applications, Vol. 33 No. 4, pp. 870-880. Wang, T.-C., Lee, H.-D. and Cheng, P.-H. (2009), “Applying fuzzy TOPSIS approach for evaluating RFID system suppliers in healthcare industry”, in Nakamatsu, K., Phillips-Wren, G., Jian, L.C. and Howlett, R.J. (Eds), New Advances in Intelligent Decision Technologies, Springer, Berlin, pp. 519-526. Wei, C.-C., Chien, C.-F. and Wang, M.-J.J. (2005), “An AHP based approach to ERP system selection”, International Journal of Production Economics, Vol. 96, pp. 47-62. Zadeh, L.A. (1965), “Fuzzy sets”, Information and Control, Vol. 8 No. 3, pp. 338-353. About the author Kazim Sari earned his PhD in Industrial Engineering from Istanbul Technical University, Istanbul, Turkey. He currently serves as Associate Professor and Chairman of International Logistics and Transportation Department at Beykent University, Istanbul, Turkey. His principal research areas include analysis and design of supply chains and logistics systems through optimization and simulation modeling. His work has been published in European Journal of Operational Research, International Journal of Production Economics, International Journal of Physical Distribution & Logistics Management and Industrial Management & Data Systems. He is also on the editorial board of International Journal of Management and Enterprise Development. Kazim Sari can be contacted at: [email protected]

To purchase reprints of this article please e-mail: [email protected] Or visit our web site for further details: www.emeraldinsight.com/reprints


465