An Efficient ANP-BGP Model for Software Production by QFD

Transcription

1 Australan Journal of Basc and Appled Scences, (0): , 20 ISSN An Effcent ANP-BGP Model for Software Producton by QFD Morteza Jamal Paghaleh Department of Industral Engneerng, Young Researchers Club, Islamc Azad Unversty, Zahedan Branch, Zahedan, Iran Abstract: In the proposed paper an effcent model va combnng QFD, Analytc Network Process (ANP), and Bounded Goal Programmng (BGP) s presented to nsure multple objectves of the organzaton and to supply a varety of customer s needs n the process of desgnng, manufacturng, and provdng the after sale servces for software products. QFD defnes customer s needs (CNs) and product s techncal requrements (PTRs) regardng ther nterrelatonshps and fnally ranks the techncal requrements. The algorthm conssts of two decson makng technques ncludng ANP and BGP. The goal of ANP s to determne and evaluate the relatonshp between CNs and PTRs. ANP approach wll facltate the evaluaton of nterrelatons between CNs and the correlaton between PTRs. By ANP, ths paper has ntroduced a BGP n order to provde the organzatons multple objectves. More over, development rate, resource constrants and the manufacturablty needed to determne PTRs were all taken nto consderatons n desgnng, manufacturng and the after sale servces of the model. The model results are demonstrated n a case study of producng software n Golestan Company to be evaluated. Key words: Analytc Network Process, Qualty Functon Development, Bounded Goal Programmng, software products. INTRODUCTION Clearly, regardng better condtons n the products and servces qualty and keepng contnuous mprovement n the process of development and change, are among major obsessons for the companes n the world of compettons. Qualty functon deployment s one of the technques whch are able to help the organzaton from the early stages of producton process.e. desgnng phase to reach the customer s satsfacton (Karask, et al., 2002; Revelle, 998; Tangm, et al., 2002). QFD dentfes the customer s needs (CNs) n a well organzed framework and reveals them all n the form of product s techncal requrements (PTRs), (Hunt and aver, 200; Shlto, 994). There has been a lot of research done over desgnng the product of QFD n the last decade n whch the man focus has always been on the customer s needs, ncludng the applcaton of fuzzy logc for regardng these demands (Chan, 990; Khoo and Hu, 996), the applcaton of analytc herarchcal process to determne the relatve mportance of the customer s needs (Lu, et al., 994; Park and Km, 998), or the applcaton of lnear programmng model n the product s desgnng process to maxmze the customer s satsfacton based on the lmts of the budget (Wasserman, 99). In the proposed artcle, a decson makng algorthm s offered whch s requred n the desgnng phase to determne and choose the product s techncal requrements. The algorthm conssts of two decson makng technques ncludng ANP and BGP. The goal of ANP s to determne and evaluate the relatonshp between CNs and PTRs. ANP approach wll facltate the evaluaton of nterrelatons between CNs and the correlaton between PTRs. As the nature of product desgnng s somehow mult objectve and t s also requred to calculate the overall prortes n techncal requrements through ANP, we should combne the results of ANP and the goal and crtera n product desgnng such as resource lmtaton, extendblty and manufacturablty by a goal programmng model; after mnmzng all the dgressons we should focus on the techncal requrements and fnally determne them. The relatve scales for the above goals are beng calculated by par comparsons and experts opnons n ths matter. Fnally, the requred techncal requrements are determned at the desgnng phase after solvng the BGP model. Ths artcle s presented n sx sectons. The second secton descrbes the tradtonal QFD framework and House of Qualty (HOQ) as are commonly dscussed n qualty management lterature. The thrd secton presents ANP fundamental and Sa'aty super matrx model. In the fourth secton, decson algorthm and dfferent phases of the proposed model are dealt wth. The model s manpulated for desgnng a software product (software selecton of Integrated Management Informaton System.e. IMIS), n secton fve. Fnally, last secton talks about the results of the proposed model. Correspondng Author: Department of Industral Engneerng, Young Researchers Club, Islamc Azad Unversty, Zahedan Branch, Zahedan, Iran. E-mal: morteza62@gmal.com 002

2 Aust. J. Basc & Appl. Sc., (0): , 20 A Revew of Qualty Functon Deployment: Qualty functon deployment was frst ntroduced by Akao n 966. It was developed at Kobe Shpyards of Mtsubsh Heavy Industres n 972. In 986 Ford Company became a poneer n usng QFD n desgned vehcle parts n the Unted State. Now t s wdely used n both Japan and the Unted States as an effectve tool n desgnng new products (Carnevall and Mguel, 2008; Erkkson and McFadden, 99; Iranmanesh and Thomson, 2008; Matook and Indulska, 2009; Shlto, 994). By usng QFD, companes wll be able to keep ther compettve advantages through three strateges of decrease n costs, ncrease n ncome, and decrease n the tme to market of the new goods and servces (Hunt and aver, 200; Khadem-Zare, et al., 200). Establshment costs, shorter crculaton n product s desgn, fewer customers complans, documentaton, determnng crtcal ponts n the product s qualty, recognzng rsk factors n the early stages of desgnng are among the other advantages of QFD (Akoa, 990; Bknel, 200; Hun, et al., 200). Further detals on the benefts of usng QFD can be found n (Carnevall and Mguel, 2008; Grffn, 992; Zare, et al., 20). Accordng to four matrces model of ASI (Amercan Suppler Insttute), QFD s a process composed of structured matrxes, based on the followng goals: a. Translatng the customer s needs to desgnng or engneerng requrements; b. Transformng mportant desgnng requrements to the parts specfcatons; c. Transformng mportant parts specfcatons to manufacturng and producton operatons; d. Transformng mportant manufacturng and producton operatons to specal operatons and ther control. The above mentoned four make a seres and the output of each wll make the nput of the next level. Researches show that less than % of the companes are successful n contnung the process beyond the HOQ (Cox, 992). Many of QFD benefts are resulted from HOQ. Ths artcle also puts ts emphass on ths matrx. House Of Qualty (HOQ): The frst phase of QFD, usually called house of qualty (HOQ), s of strategc mportance, snce t s n ths phase that the customer's needs are dentfed and then, ncorporatng the producng company's compettve prortes, converted nto approprate techncal measures to fulfll the needs. HOQ s a knd of process map whch brngs about the possblty of relatonshp between the parts (Carnevall and Mguel, 2008; Karask, et al., 2002). Snce t s so wdespread wth a varety of contents, has become the fnal pont n many real QFD projects. The dfferent phases n the creaton of HOQ are as the followngs: a. Recognzng the CNs; b. Determnng the PTRs; c. Determnng the relatve mportance of CNs; d. Determnng the relatonshp between the CNs and PTRs; f. Determnng the correlaton between the CNs; g. Determnng the correlaton between the PTRs; h. Determnng the overall prortes n PTRs and other general goals n desgnng the product (see Fg. ). The reader nterested n detals of QFD and correspondng calculatons s referred to (Grffn and Hauser, 99; Hauser and Clausng, 998; Kohen, 99; Karsak, et al., 2002; Matook and Indulska, 2009; Shlto, 994). Fg. : Schematc of a House of Qualty. 00

3 W2 0 Aust. J. Basc & Appl. Sc., (0): , 20 ANP : ANP s a mult crtera technque whch was ntroduced by (Sa'aty, 970) and dvdes the problem of decson makng nto dfferent levels whch all together form a knd of herarchy. ANP has the ablty to be used as an effectve tool n the nteractons between the elements of network structure (Carlucc and Schuma, 2009; Ergu, et al., 20; Karsak, et al., 2002; Lee and Km, 2000; Sa'aty, 999). Fg. 2 shows a network system wth feedback, nner and outer dependency. In the system there are two-way arcs, wthn the same level and among levels. Arc from component C4 to C2 ndcates the outer dependence of the elements n C2 on the elements n C4 wth respect to a common property. Loop n a component ndcates nner dependence of the elements n that component wth respect to a common property (Carlucc and Schuma, 2009) Fg. 2: Feedback Network wth clusters havng nner and outer dependence among ther elements.(carlucc, 2009) Fg. presents a sample of ANP. The symbol ( ) n the related square reveals the relatonshp between mnor goals. Generally speakng, ANP conssts of two stages: Frst s the formaton of the network and second s the factors prorty calculaton. In order to form the structure of the problem, all the nteractons should be consdered. All of these relatons and the correlaton are evaluated by usng super matrx and par wse comparsons. Super matrx s the same matrx resulted from the relatons between the network elements and the prorty vectors n these relatons (Sa'aty, 999). A super matrx for a three-phase herarchy s as the followngs: In ths super matrx W 2 s a vector showng the effect of the goal on each crteron. W 2 shows the effect of each crteron on the alternatves. The above mentoned super matrx s formed to reduce the amount of calculatons for determnng general prortes. Ths wll reveal the total effect of each factor on the other factor n ther nteracton (Sa'aty, 999). If a network contans only two groups (clusters) of crtera and solutons, then for calculatng the correlaton of the system element you can use Sa'aty or Takzava s vews presented n 986 (Karsak, et al., 2002) whch s the same method used n ths paper. Model Constructon: Desgnng the model n the proposed paper s for recognzng and choosng the product s requrements whch are requred to be consdered n the desgnng process based on the goals and restrctons. Ths decson algorthm s dvded nto two man phases; the frs phase ncludes the formaton of HOQ matrx wth the ANP approach and the second one s the combnaton of the frst phase results wth goal programmng model for determnng the product s techncal requrements needed to be taken nto consderaton by desgnng team. One of the goals of decson algorthm s the possblty of modelng and determnng the relatons between factors nsde HOQ matrx. There are some other factors such as resources restrctons, extendblty and manufacturablty whch are mportant n the analyss. Extendblty means how much of ths mprovement n techncal requrements can be transferred to other techncal requrements. Manufacturablty shows the problems n creatng or changng a techncal requrement or the efforts for the desred mprovements. For example, n order to mprove a techncal requrement one may need a specal technology, whle the other would easly be mproved. Restrctons n resources are known accordng to software s techncal specfcatons and the techncal specfcatons prorty based on the goals such as extendblty and producton capablty. Then these prortes are modfed and are calculated for determnng the correlatons n HOQ based on the goals. The relatve scales for the desred goals are determned by par wse comparsons. Fnally all of the acheved nformaton s 004

4 TR W ( Techncal Customer CN Re )' qurements sneeds W20) Aust. J. Basc & Appl. Sc., (0): , 20 combned wth a BGP model for determnng techncal specfcatons n desgnng process. Fg.4 shows ths algorthm step by step. Fg. : A sample of ANP. Snce goal programmng wth lmted and balanced equvalents can consder multple objectves wth ther relatve mportance and mnmze the overall devaton from these goals, therefore t s preferable to use ths knd of programmng n ths research as a decson makng tool. Ths specal feature can pave the way to enter some goals ncludng restrcton n resources, extendblty and manufacturablty n the product s desgnng process. Goal programmng mnmzes the overall balance devaton from the mentoned goals and studes them at the same tme. The scales are not ranked but can reflect the decson maker s prortes accordng to the relatve mportance of each goal. Lack of balance n scales, happenng n BGP durng measurng goals wth dfferent unts, s solved by normalzaton (Schnederjans & Garvn, 997). The super matrx showng QFD model n ths artcle s as followngs: Step. Recognzng CN and ther related TR Step2.Studyng the correlatons and relatons n HOQ and determnng TRs overall prortes by ANP Step. Recognzng measurement crtera and the resources restrctons Step 4. Determnng TRs preferences based on the desgnng goals Step. Modfyng restrctons n resources and other desgnng goals Step6. Calculatng the relatve weghts of desred goals by par wse comparsons Step7. Formulatng BGP model and solvng t n order to determnng the TRs of desgnng phase Fg. 4: Step by step process of the proposed decson algorthm. W shows the effects of the goal on customer s needs; n other words t shows the mportance of CNs. W 2 s a matrx showng the effect of customer s needs on each techncal requrement. W s the matrx showng the nterrelatons n CNs and W 4 s the matrx showng nterrelatons n PTRs. QFD model network s shown n Fg.. Accordng to the above abbrevatons, the nterrelatons of CNs (W ) multpled by the relatve 00

5 Aust. J. Basc & Appl. Sc., (0): , 20 mportance of CNs (W ) equals to the prortes of the CNs correlaton. Exactly the same as what was mentoned above, the nterrelaton of the PTRs (W 4 ) multpled by the relatonshp of CNs wth each of the PTRs (W 2 ) equals to prores of TRs correlaton. After calculatng W CNs and W PTRs, the general prortes for techncal requrements shown by W ANP, are calculated as the followngs: W CNs =W * W W PTRs =W 2 * W 4 W ANP =W PTRs * W CNs Achevng the Best Product Desgn CNs PTRs Fg. : Network model of QFD. Based on a group of techncal requrements whch should be taken nto consderaton n desgnng phase, a BGP model s made wth the frst levels results. Ths model s also related to the goals whch we should focus on durng the desgnng perod ncludng restrctons n resources, extendblty, and manufacturablty. The goals related to the resources restrctons are easly formulated. But other goals such as extendblty and manufacturablty wll result n preferental scales for the techncal requrements. In order to formulze these goals n the model, we need to determne a preferental rate for each requrement and then the negatve dervatons of these goals from one, are appled n the goal functon. Goal programmng model wth lmted varatons: In order to meet the organzaton and customer goals and restrctons smultaneously, we are usng bounded goal programmng n ths artcle. The general form of ths model n desgn makng algorthm s as the followngs: MnZ s. t. n j n n j W r j j W j j 0 0 W d d d d d d d d d, d,2,..., m j j j j m R 2,,..., m 2,,..., m j,2,..., n W s the goals weght n the balanced goal model. d + and d - shows the varables of postve and negatve devatons n the th goal. x j, lmted varable between zero and one, shows the jth PTR. W j shows the relaton prorty of the jth PTR. r j s a pat of the th sources used by the jth PTR. R shows resource restrcton n th resource, and w j s the th relatve mportance n PTRs based on the th goal or crteron. Model Employment: In ths secton the decson algorthm and the model makng process wll be dscussed based on a case study on the software of the comprehensve system n management data.the par wse comparson tables and some relatve matrxes are not mentoned here and we just put the matrx results. The results of par wse comparsons used here, are all the outcome of the experts and managers opnons n QFD team whch are taken to manpulate the model n the software company. In fact, QFD team s the statstcal group of decson makng. 006

6 Aust. J. Basc & Appl. Sc., (0): , 20 Step : In our case study, 6 CNs were determned ncludng data securty, relablty, accessblty, speed, proper reacton toward users mstakes and the possblty of removng the dffculty. After determnng the CNs, by the help of system desgners, software experts and the exstng standards for software qualty ncludng CMM (Capablty Maturty Model) and ISO-926, ten PTRs were determned for the software, ncludng the average nterval between defects ( ), the average tme for repar ( 2 ), the rejecton amount of nconsstent data ( ), accessblty of shared data ( 4 ), the average tme for data retreval ( ), the possblty of takng backup fles ( 6 ), program access and control level ( 7 ), the securty level of data accessblty ( 8 ), the average tme of data changng ( 9 ), executng exe fles ( 0 ). Step 2: In ths step we determne the nterrelatons between CNs and PTRs through the prevous step data and the ntervews done wth experts and the techncal crtera. Ths step can be dvded nto 4 parts: Determnng the relatve mportance of CNs: Suppose that there s ndependency between CNs, so n order to determne ther mportance we can ask ths queston Whch customer s needs are more mportant? the normalzed matrx out of the group opnon.e. W s as the followng: W CN 0.86 CN CN 0.6 CN4 0.7 CN 0.6 CN Determnng the relatonshp between CNs and PTRs: In order to determne the relaton between the CN nput securty and PTRs, we should ask ths queston Accordng to the customer s need, whch techncal requrement s more mportant, the rejecton amount of ncompatble data or the possblty of takng back up fles? ; If we do the same thng about the other needs and requrements, the fnal matrx showng the relatve scales for the software techncal requrements n accordance to each customer s needs (W 2 ) wll be as the followngs: CN CN2 CN CN4 CN CN W Calculatng the nner dependence (correlaton) of the customer s needs : Based on the relatons between CNs, llustrated n Fg.6, the effect of each need on others s calculated through par wse comparsons. For example you can ask ths queston based on the data securty that s supposed as the major CN, how much s the mportance of other CNs? ; so the major CN weght would equal to one, those of unrelated CNs wth data securty would equal to zero and the rest CNs' weghts would be between one and zero regardng to ther correlaton wth data securty. The group judgment results and the normalzed vector of data securty mportance comparng to other CNs s cted n Table. Fg. 6:Correlaton between CNs. 007

7 Aust. J. Basc & Appl. Sc., (0): , 20 Table : CNs prortes and weghts. CN CN Prorty CN Weght CN 0.74 CN CN CN CN CN Sum.4 Table 2: Requred man-hour for each PTR (hr) PTR Man-Hour In case that we follow the same procedure for the other needs, the knd of matrx showng the correlaton between CNs (W ) s as the followng: Calculatng the correlaton of the software Techncal Requrements: In order to calculate ths correlaton, the network structure llustrated n Fg.7 s used. The correlaton matrx between PTRs (W 4 ) s as the followng: Fg. 7: Correlaton between PTRs. x x2 x x4 x x6 x7 x8 x9 x W The overall relatve weghts of CNs based on the correlatons.e. WCNs, are as the followng: CN 0.8 CN CN 0.2 WCNs W W CN CN 0.90 CN

8 Aust. J. Basc & Appl. Sc., (0): , 20 The relatve weghts of PTRs based on the correlatons.e. W PTRs, are as the followngs: CN CN2 CN CN4 CN CN WPTRs W4 W Fnally, the overall relatve weghts of PTRs (W PTRs ) based on the overall relatve weghts of CNs (W CNs ).e. W ANP whch reflect the overall relatons nsde HOQ (Fg.7), are as the followng: W ANP W PTRs W CNs The results of ANP analyss show that the most mportant requrement n software desgnng for IMIS s the rejecton of nconsstent data ( ) wth the relatve weghts of On the second prorty, there s the program access and control level ( 7 ), wth the relatve weghts of Step : In ths step the crtera related to the restrcton of resources come nto the decson makng process. The estmated man-hour requred to create or correct any of PTRs s accordng to Table 2. On the other hand, the total requred man-hour that the company can allocate for software update or development was 600 man-hour based on the number of experts and ther workng hours. Step 4: There are other goals, regardless of budget and the workforce, n software desgnng ncludng extendblty and manufacturablty of PTRs, whch are determned on a fve-level scale from very low to very hgh. If the results are normalzed, we wll have W M and W E matrxes as bellows: W M, WE Step : The vectors related to resources restrctons, extendblty and manufacturablty should reflect the correlatons and nterrelatons n the software s techncal requrements. h ' : The adjusted vector for the requred man-hourof PTRs. W E ' : The adjusted vector for PTRs' extendblty. W M ' : The adjusted vector for PTRs' manufacturablty. 009

9 Aust. J. Basc & Appl. Sc., (0): , 20 W ' W W E 4 E h' W4 h W ' W W M 4 M Now wth all the acheved data, we are able to buld the HOQ. Fg.8 shows the HOQ related to the process. Step 6: In desgnng a software or any other product, there are always some goals and crtera at the same tme. The goals and crtera studed n ths paper n desgnng the software are: focusng on the nterrelatons between each group of factors wth ANP, the avalable workforce, extendblty and manufacturablty. In order to measure the relatve weghts of these crtera by par wse comparsons we are n need of expert vews whch have been normalzed as the followngs (W j ): W j ANP Workforce Extensblty Manufacturablty Fg. 8: The HOQ resulted from step to. Step 7: We can buld the goal programmng model wth the balanced bounded varables, requred n determnng the software s TRs durng desgnng level, by applyng our completed HOQ (Fg.8), the relatve weghts of the desgnng goals and crtera (W j ) and BGP., 2,, 0, show decson makng varables or the software s TRs n BGP model. The frst restrcton s related to the overall prortes of PTRs. The thrd s the average amount of extendblty, and the forth s related to manufacturablty of the software s TRs. The rght hand sde of the second constrant ndcates the avalable man-hour. The rght hand sde of the frst, thrd and forth constrants are related to lateral scales and prortes of TRs, whch are at most equal to one. The objectve functon s to mnmze the postve and negatve devatons from the desred value. 00

10 Aust. J. Basc & Appl. Sc., (0): , 20 MnZ 0.6(d d ) 0.04(d2 d 2 ) 0.094(d d ) 0.284(d4 d 4 ) s.t () d d (2) d2 d () d d (4) d4 d4 0 j j,2,...,0 0 d,d,2,,4 After solvng BGP model by Lngo software, the followng results are derved: d 0.07 d d 0. d d d2 d d4 0 As can be seen,, 2,, 6, 7 and 8 are equal to, whle 4, and 9 are equal to 0.9 and 0 s equal to 0.8. It means that the software s TRs such as breakdown average tme, the average tme n repars, the amount of rejectng nconsstent data and the possblty of takng a backup fle are all chosen for full securng. But other requrements ncludng shared data avalablty the average tme n data retreval the average tme n changng data and the tme needed n openng a fle are all chosen for partal securty. The amounts of d, d 2, d, and d 4 show undesrable devatons from goals and ANP crtera. Ths knd of devaton from the desred soluton s due to nconsstency and contradctons n organzatonal and customer goals. Concluson: As all other tools, the advantages and effectveness of QFD depend on how t s beng manpulated. In order to make t more effectve, n the proposed paper, a knd of systematc approach of decson makng was ntroduced n desgnng process of the product n QFD. Ths decson algorthm facltates the study of the relatons between customer s needs, techncal requrements, correlatons between them, applyng resource restrctons and also other crtera ncludng extendblty and manufacturablty. In ths era of ncreasng compettveness, we should meet the nteractons between dfferent deas and use them to reach a combned model n QFD. Therefore, the potentals of QFD as an effectve tool wll become more actual and possble. In ths respect, ths paper has tred to apply the combned model of ANP and BPG for nterferng customer s needs and product s techncal requrements n product desgnng by QFD. The correlaton of dfferent elements n QFD based on ANP was appled n the decson algorthm. Accordng to the restrctons n resources and the mult-objectvty of the problem, a BPG model was made n order to determne the techncal requrements and the product s controllng ponts for customer s satsfacton. We should bear n mnd that QFD alone can not be a tool for optmum desgnng; the prortes resulted from ANP, restrctons n resources and other crtera n desgnng such as extendblty and manufacturablty n BGP model gve more justfable answers. Also, the desgnng necesstes are determned n a way that maxmzes customer s satsfacton based on the exstng restrctons. REFERENCE Akao, Y., 990. Qualty functon deployment: Integratng customer requrements nto product desgn. Cambrdge, MA: Productvty Press. Bcknell, B.A., K.D. Bcknell, 99. The Road Map to Repeatable Success Usng QFD to Implement Change. CRC Press, Boca Raton, FL. Carlucc, D., G. Schuma, Applyng the analytc network process to dsclose knowledge assets value creaton dynamcs. Expert Systems wth Applcatons, 6(4): Carnevall, J.A., C. Mguel, Revew, analyss and classfcaton of the lterature on QFD-Types of research, dffcultes and benefts. Internatonal Journal of Producton Economcs, 4(2): Chan, L.K., H.P. Kao, A. Ng, M.L. Wu, 999. Ratng the mportance of customer needs n qualty functon deployment by fuzzy and entropy methods. Internatonal Journal of Producton Research, 7(): Chan, L.K., M.L. Wu, Qualty functon deployment: A lterature revew. European Journal of Operatonal Research, 4:

11 Aust. J. Basc & Appl. Sc., (0): , 20 Chan, L.K. andm.l. Wue, 200. A systematc approach to qualty functon deployment wth a full llustratve example. Omega, : 9-9. Cohen, L., 99. Qualty functon deployment: How to make QFD work for you. Readng, MA: Addson- Wesley. Cox, C.A., 992. Keys to success n qualty Functon Deployment. APICS The Performance Advantage pp: Ergu, D., G. Kou, Y. Sh and Y. Sh, 20. Analytc network process n rsk assessment and decson analyss. Computers and Operatons Research. In Press, Accepted Manuscrpt, Avalable onlne 29 March 20. Erkkson, I. and F. McFadden, 99. Qualty functon deployment: a tool to mprove software qualty. Informaton and Software Technology, (9): Grffn, A., 992. Evaluatng QFD s use n US frms as a process for developng products. Journal of Product Innovaton Management, 9(): Grffn, A. and J.R. Hauser, 99. The voce of customer. Marketng Scence, 2(): -27. Hauser, J.R. and D. Clausng, 988. The House of Qualty. Harvard Busness Revew, pp: 6-7. Hun, S.B., S.K. Chen, M. Ebrahmpour, and M.S. Sodh, 200. A conceptual QFD plannng model. Internatonal Journal of Qualty and Relablty Management, 8(8): Hunt, R.A. andf.b. aver, 200. The leadng edge n strategc QFD. Internatonal Journal of Qualty & Relablty Management 20(), 6-7. Iranmanesh, H. and V. Thomson, Compettve advantage by adjustng desgn characterstcs to satsfy cost targets. Internatonal Journal of Producton Economcs, : Karsak, E.E., S. Sozer and S. Emre Alptekn, Product plannng n qualty functon deployment usng a combned analytc network process and goal programmng approach. Computers and Industral Engneerng, 44(): Khadem-Zare, H., M. Zare, A. Sadegheh and M.S. Owla, 200. Rankng the strategc actons of Iran moble cellular telecommuncaton usng two models of fuzzy QFD. Telecommuncatons Polcy, 4: Lee, J.W. and S.H. Km, Usng analytc network process and goal programmng for nterdependent nformaton system project selecton. Computers and Operatons Research, 27(4): Matook, S. and M. Indulska, Improvng the qualty of process reference models: A qualty functon deployment-based approach. Decson Support Systems, 47(): Park, T. and K. Km, 998. Determnaton of an optmal set of desgn requrements usng house of qualty. Journal of Operatons Management, 6: Revelle, J.B., J.W. Moran and C.A. Cox, 998. The QFD Handbook. Wley, New York. Sa'aty, T.L., 999. Fundamental of the Analytc Network Process. Berne, Swtzerland: ISAHP Conference Presentaton. Schnederjans, M.J., T. Garvn, 997. Usng the analytc herarchy process and mult-objectve programmng for the selecton of cost drvers n actvty-based costng. European Journal of OperatonalResearch, 00: Shllto, M.L., 994. Advanced QFD -Lnkng technology to market and company needs. New York: Wley. Tang, J., R.Y.K. Fung, B. u and D. Wang, A new approach to qualty functon deployment plannng wth fnancal consderaton. Computers and Operatons Research, 29(): Zare, M., M.B. Fakhrzad and M. Jamal Paghaleh, 20. Food supply chan leanness usng a developed QFD model. Journal of Food Engneerng, 02: