International Journal of Scientific & Engineering Research Volume 2, Issue 7, July-2011 1

ISSN 2229-5518

An Integrated Approach of AHP-GP and Visualization for Software Architecture Optimization: A case-study for selection of architecture style

K.Delhi Babu, P.Govindarajulu, A.Ramamohana Reddy, A.N.Aruna Kumari

Abstract— Software Architecture has emerged as an important sub-discipline of software engineering. A key aspect of the design of any software is its architecture styles, i.e. components and connectors and their relationships. Selecting the best style is difficult because there are multiple factors such as project risk, corporate goals, limited availability of resources, etc. Therefore this study presents a methodology for selection of software architecture styles. In this paper we explore the Analytic Hierarchy Process (AHP) within a zero-one goal programming (ZOGP) model for selection of architecture styles. AHP is applied to the decision problem involving multiple alternatives and criteria and aims at selecting an alternative from a known set of alternatives. Then Goal programming model is used to optimize the objective function while simultaneously satisfying all the constraints. Further, AHP-GP Visualization framework and visualization tool (SAVE Tool) are applied to evaluate the selected software architecture style.

Index Terms— Software Architecture, Selection of Software Architecture Styles, Multi-Criteria Decision Making, Analytic

Hierarchy Process (AHP), Zero-One Goal Programming (ZOGP), Visualization.

—————————— • ——————————

1 INTRODUCTION

owadays, decision-making has become more com- plex due to reasons related to (1) the alternatives, (2) the goals and (3) the environment in which deci-
sions are being made. First, for almost any decision the number of alternatives has grown dramatically. Second, the number and the nature of the goals, criteria or con- straints, have changed. When making decisions, the goals are not limited to related objectives. The third set of rea- sons for the increased complexity of decisions refers to the environment. The changing alternatives, goals, and environment enlarge the complexity of decisions and call for effective decision support. The basic approach of ma- thematical programming models is to optimize the objec- tive function while simultaneously satisfying all the con- straints that limit the activities of the decision maker.
Software architectures significantly impact software project success [1]. However, creating architectures is one of the most complex activities during software develop- ment [2]. When creating architectures, architecture styles narrow the solution space: First, styles define what ele- ments can exist in architecture (e.g. components, connec- tors). Second, they define rules on how to integrate these

————————————————

K.Delhi Babu, Department of Computer Science and Engineering, Sri Vidyanikethan Engg. College, Tirupati, Andhra Pradesh, INDIA. Email:kdb_babu@yahoo.com

P.Govindarajulu, Department Computer Science, S.V. University, Tirupa-

ti, Andhra Pradesh, INDIA. Email: pgovindarajulu@yahoo.com

A.Ramamohana Reddy, Department Computer Science, S.V. University,

Tirupati, Andhra Pradesh, INDIA. Email: aramamohanareddy@yahoo.com

A.N.Aruna Kumari, Department of Computer Science and Engineering,

Sri Vidyanikethan Engg. College, Tirupati, Andhra Pradesh, INDIA. Email: kolla_aruna@yahoo.com
elements in the architecture. Moreover, styles address func- tional and non-functional issues [3]. This paper focuses on two mathematical methods, Analytic Hierarchy Process and Goal Programming. Further, AHP-GP-Visualization framework and Visualization Tool (SAVE Tool) are applied to evaluate the selected software architecture style.
The contributions of this paper are as follows:
1. This paper presents a methodology for selection of
software architecture style which uses two mathe-
matical techniques Analytic Hierarchy Process and
Goal Programming.
2. Analytic Hierarchy Process (AHP) is used to deter-
mine the degree of relative importance among the al-
ternatives and criteria.
3. It provides a way of collecting expert group opinion
along with stakeholders interests (e.g. reliability, per-
formance)
4. Goal Programming (GP) to determine the desired
level of attainment for each goal and penalty weights
for over or under achievement of each goal [4]
5. AHP-GP Visualization Framework and Visualization
Tool are used to evaluate the selected software archi-
tecture style.
In AHP, pairwise comparisons matrix is formulated and then the relative priority of each alternative is calcu- lated. After obtaining the overall priorities of alternatives and using the goal constraints, zero-one goal program- ming (ZOGP) model formulated. The combined use of the AHP and GP approaches extended the use of Multi Crite- ria Decision-Making approach.

IJSER © 2011 http://www.ijser.org

International Journal of Scientific & Engineering Research Volume 2, Issue 7, July-2011 2

ISSN 2229-5518

2 INTEGRATED APPROACH OF AHP-GP AND

VISUALIZATION

2.1 Analytic Hierarchy Process

The initial study identified the multi-criteria decision technique known as the Analytic Hierarchy Process (AHP) to be the most appropriate for solving complicated problems. Decision-making involving multiple objectives

2.2 Goal programming

After obtaining the overall priorities of alternatives using AHP, with these priorities and goal constraints, Goal Programming (GP) model is formulated [9] as illu- strated in figure 1, step 5 to step 6.
The GP model for architecture style selection can be stated as follows:
and/or criteria is called Multi Criteria Decision-Making
Minimize

Z = P

(w d + , w d - )

(1)
(MCDM) [5]. Often the criteria include both qualitative

K j i j i

and quantitative factors, whereas the quantitative criteria may be measured in incomparable units (for example, the
Subject to

aij x j + di

- di

bi

market share and the price of a software package). T.L. Saaty introduced AHP to solve the problem of indepen- dence on alternatives and/or criteria. AHP allows better, easier and more efficient identification of selection crite- ria, their weighting and analysis [6]. It reduces drastically the decision cycle and allows organization to minimize
common mistakes by using the expert group decision [7].
for i = 1, 2, …, m, j = 1, 2, …, n (2)

x +d- =1

for i = m+1, m+2, …m+n, j = 1, 2, …, n (3)

x j = 0 or 1 for Vj

(4)

Figure 1. Process Model for architecture style selection using

AHP-GP Visualization

Thus, AHP is “a method of breaking down a com- plex, unstructured situation into its components parts; arranging these parts, or judgments on the relative impor- tance of each variable; and synthesizing the judgments to determine which variables have the highest priority and should be acted upon to influence the outcome of the sit- uation” [8]. In Analytic Hierarchy Process, a first pairwise comparison matrix for alternatives and criteria are formu- lated and then the relative priority of each alternative and criterion is calculated. In Figure 1, step 1 to step 4 depict AHP model.
where m is the number of goals to be considered in the model, n is the pool of architecture styles from which the optimal set will be selected, = the AHP mathematical weight on the j =1, 2,…, n architecture style, = some k priority pre-emptive priority , for i =1, 2,…, m goals, = the ith positive and negative deviation variables for i = 1,
2,…, m goals, = a zero-one variable, where j = 1, 2,…, n possible projects to choose from and where = 1, then se- lect the jth architecture style or when =0, then do not se- lect the jth architecture style, = the jth parameter of the ith resources, and = the ith available resource or limita- tion factors that must be considered in the selection deci- sion. In Figure 1, steps 5 to step 7 depict GP model.
The presented GP formulation can easily be rear- ranged or modified depending on the priorities of the decision makers and circumstances of the decision envi- ronment. The GP objective function includes the positive and negative deviational variables which represent the deviations from the desired goal levels (i.e., over- achievement of a goal is represented by d+ and undera- chievement of a goal is shown as d-). Resource limitations are considered more important. The solution of the GP model will minimize the objective function and satisfy the goal constraints.

2.3 AHP-GP Visualization Framework for

Architecture Style Selection

Visualization is used to enhance information under- standing by reducing cognitive overload. The proposed AHP-GP Visualization Model as shown in Figure 2, over- come the limitations of the AHP-GP model as AHP-GP model address the non-functional requirements of the software architecture as desired by the stakeholder. The visualization techniques address the functional key areas only.

IJSER © 2011 http://www.ijser.org

International Journal of Scientific & Engineering Research Volume 2, Issue 7, July-2011 3

ISSN 2229-5518


The proposed model is the best model to choose ar- chitecture style suitable for a given application as per both the functional and non-functional requirements [10] as in figure 1, step 9.
Figure 2. AHP-GP-Visualization Framework for the selection of architecture style [10]

2.4 Evaluating the Architecture Style using

Visualization Tool (SAVE Tool)

Fraunhofer SAVE [11] (Software Architecture Visua- lization and Evaluation) is a tool for analyzing and opti- mizing the architecture of implemented software sys- tems. Using this tool the functional criteria as specified in the AHP-GP-Visualization Framework can be evaluated effectively as depicted in figure 1, step 10. Even from the source code, the architecture styles used in the project can be visualized and the functional criteria can be evaluated. Also SAVE supports reverse engineering, quality assur- ance, and maintenance tasks for systems implemented in Java, C/C++ and the extracted information can be visua- lized, analyzed, manipulated or used to modify system artifacts.

3 A CASE-STUDY FOR SELECTION OF SOFTWARE

ARCHITECTURE STYLE

A case study to illustrate the advantages of the inte- grated AHP-GP based on the expert opinion of an organi- zation is taken. The problem consisted of prioritizing three architectures styles [12] on the basis of seven criteria deemed to be important for an organization. The criteria used are (1) Modifiability (M), (2) Scalability (S), (3) Per- formance (P), (4) Cost (C), (5) Effort (E), (6) Portability (Pr) and (7) Ease-of-use (Eu). However, we are of the opi- nion that there is an existence of relative importance among these seven criteria. The attribute of criteria P in- fluence criteria C, the attribute of criteria E influence cri- teria Eu, S, Pt, C and so on. In order to check relationship of criteria or alternative, we need to have group discus- sion because the type of network or relationship depends on the stakeholders' judgment.

3.1 Decision Hierarchy Formulation

The decision hierarchy formulation is very important as the export group agreed that the evaluation criteria in the decision criteria are comprehensive and also agreed that the criteria should be expressed in fairly general terms and should be understood by all stakeholders. They concerned both managerial and technical factors were critical decision criteria. Based on the functional and non- functional requirements, stakeholders and export group opinion and previous project information are considered to in the formulation of decision hierarchy [13].

Figure 3. AHP-GP-Visualization Decision Hierarchy for the selection of architecture style

3.2. Pairwise Comparisons

This section focuses on the comparison of the alterna- tives, with respect to the other alternatives in the hie- rarchy. The judgment of the importance of one of the al- ternative over the other can be made subjectively. The subjective judgment that is achieved can then be con- verted to a numerical value using a satty scale of 1-9 , where 1 denotes equal importance and 9 denotes the highest degree of importance. For this pair wise compari- son, we follow bottom up method. In this pair wise com- parison process, all the obtained comparison results are evaluated by the expert group to better reflect their per- ception and understanding of the issues.

3.3. Decision Weight calculation

This is the main step in the selection procedure. This step focuses on getting the input from the Expert’s group, i.e., comparison of matrices. Then the relative weights for the available alternatives with respect to the available criteria are calculated.

IJSER © 2011 http://www.ijser.org

International Journal of Scientific & Engineering Research Volume 2, Issue 7, July-2011 4

ISSN 2229-5518

1. Pair wise comparison with respect to Modifiability

Modifiability PF LA BB PF 1 3 5

LA 1/3 1 3

BB 1/5 1/3 1

PF 0.6521 0.6923 0.5555

LA 0.2174 0.2307 0.3333

BB 0.1304 0.0769 0.1111

Relative Priorities for Modifiability is w11 (0.6333, 0.2604, 0.1061)

2. Pair wise comparison with respect to Scalability

Scalability PF LA BB PF 1 3 7

LA 1/3 1 3

BB 1/7 1/3 1

PF 0.6774 0.6923 0.6363

LA 0.2258 0.2307 0.2727

BB 0.1428 0.0769 0.0909

Relative Priorities for Scalability is w12 (0.6686, 0.2431, 0.1035)

3. Pair wise comparison with respect to Performance
Pair wise comparison with respect to Probability

Probability PF LA BB PF 1 1/7 3

LA 7 1 1/5

BB 1/3 5 1

PF 0.1200 0.0232 0.7142

LA 0.8400 0.1627 0.0476

BB 0.0400 0.8139 0.2380

Relative Priorities for Probability is w16 (0.2858, 0.3501, 0.3640)

6. Pair wise comparison with respect to Ease of Use

Ease of Use PF LA BB PF 1 7 3

LA 1/7 1 7

BB 1/3 1/7 1

PF 0.6774 0.8596 0.2727

LA 0.0967 0.1228 0.6363

BB 0.2258 0.0175 0.0909

Relative Priorities for Ease of Use is w17 (0.6032, 0.2853, 0.1114)

Relative Importance Matrix Calculations for the above

stated criteria:

Performance PF LA BB

PF 1 7 5

LA 1/7 1 3

BB 1/5 1/3 1

PF 0.7446 0.8400 0.5556

LA 0.1063 0.1200 0.3333

BB 0.1489 0.0400 0.1111

Relative Priorities for Performance is w13 (0.7134, 0.1865, 0.1000)

4. Pair wise comparison with respect to Cost

Cost PF LA BB PF 1 3 5

LA 1/3 1 7

BB 1/5 1/7 1

PF 0.6521 0.7241 0.3846

LA 0.2173 0.2413 0.5384

BB 0.1304 0.0344 0.0769

Relative Priorities for Criterion Cost is w14 (0.5869, 0.3324, 0.0806)


5. Pair wise comparison with respect to Dev. Effort

Pair wise comparison for Criteria

W M S P C E Pr Eu

M 1 3 3 5 7 3 7

S 1/3 1 3 5 3 7 3

P 1/3 1/3 1 3 5 5 7

C 1/5 1/5 1/3 1 5 7 3

E 1/7 1/3 1/5 1/5 1 7 5

Pr 1/3 1/7 1/5 1/7 1/7 1 3

Eu 1/7 1/3 1/7 1/3 1/5 1/3 1

Normalized Pair-wise Comparison Matrix:

W M S P C E Pr Eu

M 0.4022 0.5614 0.3808 0.3406 0.3279 0.0989 0.2413

S 0.1340 0.1871 0.3808 0.3406 0.1405 0.2307 0.1034

P 0.1340 0.0623 0.1269 0.2044 0.2342 0.1648 0.2413

C 0.0804 0.0124 0.0423 0.0681 0.2342 0.2307 0.1034

E 0.0574 0.0623 0.0084 0.0136 0.0468 0.2307 0.1724

Pr 0.1340 0.0017 0.0084 0.0019 0.0066 0.0329 0.1034

Eu 0.0574 0.0623 0.0012 0.0006 0.0013 0.0109 0.0344

Relative Importance Matrix:

Development

Effort

PF LA BB

W M S P C E Pr Eu RowAvg

PF 1 3 7

LA 1/3 1 3

BB 1/7 1/3 1

PF 0.6774 0.7142 0.5384

LA 0.2258 0.2380 0.3846

BB 0.0967 0.0476 0.0769

Relative Priorities for Dev. Effort is w15 (0.6433, 0.2828, 0.0737)

M 0.4022+0.5614+0.3808+0.3406+0.3279+0.0989+0.2413/7 0.3361

S 0.1340+0.1871+0.3808+0.3406+0.1405+0.2307+0.1034/7 0.2167

P 0.1340+0.0623+0.1269+0.2044+0.2342+0.1648+0.2413/7 0.1668

C 0.0804+0.0124+0.0423+0.0681+0.2342+0.2307+0.1034/7 0.1102

E 0.0574+0.0623+0.0084+0.0136+0.0468+0.2307+0.1724/7 0.0845

Pr 0.1340+0.0017+0.0084+0.0019+0.0066+0.0329+0.1034/7 0.0412

Eu 0.0574+0.0623+0.0012+0.0006+0.0013+0.0109+0.0344/7 0.0240

IJSER © 2011 http://www.ijser.org

International Journal of Scientific & Engineering Research Volume 2, Issue 7, July-2011 5

ISSN 2229-5518

1.3. Weight Aggregation

After all the relative weights are calculated, a composite weight for each decision choice is determined by aggregating

the weights over the hierarchy for decision choice.

Table. Cost and resources usage information

Project resource usage ( aij )

x1 x2 x3 bi

Calculation of Relative Priorities:

T0.3361 0.3361 0.3361

Planning and design days

10 24 18 24 days

T0.6333

0.6686

0.7134

0.5869

0.6433

0.2858

0.6032

0.2167

0.1668

0.2167

0.1668

0.2167

0.1668

Construction months 32 34 30 35 months


Budgeted cost (00) $150 $300 $280 $300

0.2604

0.2431 0.1865

0.3324

0.2828

0.3501 0.2853 X 0.1102

0.1102

0.1102

Misc cost (00) $18 $24 $15 $42

L0.1061

0.1035

0.1000

0.0806

0.0737

0.3640

0.1114 J

0.0845

0.0412

0.0845

0.0412

0.0845

0.0412

T0.6220

= 0.2531

L0.1075

0.6220

0.2531

0.1075

0.6220

0.2531

0.1075 J

0.0240 0.0240

0.0240

Based on the weight vector computed using AHP, we can formulate the goal constraints in the following table. This ZOGP model is solved using LINDO Ver 6.1. The re- sults are summarized as follows:

Table. ZOGP model formulation

ZOGP model formulation Goals

Normalized Priority Matrix:

Minimize Z =

pl d + + d + + d +

Satisfy all obligatory goals

T0.6220

0.6220

0.6220

1 ( 1 2 3 )

- -

- Select highest AHP weighted

0.2531

0.2531

0.2531

pl (0.622d

+ 0.253d

+ 0.101d )

L0.1075

0.1075

0.1075 J

2 5 6

7 architecture styles

Use $20,000 for all architec-

pl3 (d 8 - + d 8 + )

ture styles selected

pl d - + d +

Use $4200for all architecture

Relative Priority Vector:

4 ( 4 4 )

styles selected

Subject to

Pipes & Filters : 0.6220+0.6220+0.6220/3 = 0.622

10 X

+ 24 X

+ 18 X

- +

+ d - d = 24

Avoid over utilizing max.

Layered Style : 0.2531+0.2531+0.2531/3 = 0.253

1 2 3 1 1

planning and design days

32X + 34X

+ 30X

+ d - - d + = 35

Avoid over utilizing max.

Black Board : 0.1075+0.1075+0.1075/3 = 0.101

1 2 3 2 2

- +

construction months

Avoid over utilizing max.

150X +300X +280X +d -d

= 300

The final results obtained in the AHP Phase are (PF, LS, BB) = (0.622, 0.253, 0.101). These weights are used as priori- ties in goal programming formulation. That is (PF, LS, BB) = ( w1 , w2 , w3 ) = (0.622, 0.253, 0.101) are the values of the three

architecture styles.

1 2 3 3 3

X 1 + d 5 - = 1

X 2 + d 6 - = 1

X 3 + d 7 - = 1

- +

budgeted dollars

Select Layered Style (LS) Select Pipe & Filter (PF)

Select Blackboard Style (BB)

X X X d d

Avoid over or under utilizing

The weight vector obtained from the above AHP model is

18 +24 +15 + -

1 2 3 4 4

=42

misc cost

used to optimize the solution further by zero-one goal pro-

150X + 300X

- +

+ 280X + d - d = 300.

Avoid over or under utilizing

gramming as follows: There exist several obligatory and flex-

1 2 3 8 8

expected budget

ible goals that must be considered in the selection from the available pool of three architecture styles. There are three obligatory goals: (1) a maximum time of 24 working days is

Xj = 0 or Vj = 1,2,3

required to select the best architecture style, (2) a maximum

x1 = 0 x2 = 1,

x3 = 0

duration of 35 months is required to complete the software

d - = 0,

d + = 0,

d - =1,

d + = 0,

d - = 0,

d + = 0,

d - =18,

d + = 0,

project and (3) a maximum budget of $ 20,000 is allocated to

develop the project [14]. In addition to the obligatory goals of selecting the best architecture style, there are two other flexi-

d - = 1,

d - = 0,

d - = 1,

d - = 0,

d + = 0.

ble goals, stated in order of importance: (1) allocation of budget is set at $20,000 and (2) allocation of miscellaneous fees is set at $4200, deviation from this allocation is not al- lowed. In the following table, the cost and resource usage information for each of the three styles is presented.

Architecture Style Pipe & Filter (PF) is chosen as it con-

sumes the total budgeted cost of $30,000 and use 24 days of time for decision. Also, the selected style will save one month construction time (total time is 35 months) as . Com- pliance checking is to be executed for every single modifica- tion made to source code using the SAVE Tool.

IJSER © 2011 http://www.ijser.org

International Journal of Scientific & Engineering Research Volume 2, Issue 7, July-2011 6

ISSN 2229-5518

4 DISCUSSION

As organizations are infusing more technologies to sup- port business operations and enhance organizational compe- titiveness, they are like to find that the traditional methods such as scoring, ranking methods, Fuzzy logic, etc analyses are unsatisfactory for evaluating emerging technologies for adoption of decisions.

The objective of this paper is to provide a systematic process to sort out the available alternatives by applying the decision analysis model, the AHP. However, the comparison of the AHP and other method like scoring and ranking me-

The proposed model, AHP-GP is to demonstrate the proce- dure of finding weight that considers interdependence among criteria or alternatives [20] which has highest weight wj. The ZOGP model selects the best architectural style for which the weight wj is derived from AHP which has maxi- mum value and minimum deviation dj. Finally, architecture Style 2 is chosen which is optimum as it is consumes the total budget cost of $30,000 and use exactly 24 days of time for decision. The selected style will save one month construction

-

thods may be inappropriate as the AHP is a multi criteria

decision technique. Another extension of this research could test the wide range of decision problem is a group setting. This study of AHP has evaluated in the context of Software

time (total time is 35 months) as

5 CONCLUSIONS

d 2 = 1

Architecture selection. A more comprehensive list of multi criteria problems, including varied level of difficulty and problem types, should be prepared to guide the systematic evaluation of the strengths and weakness of AHP. It would perhaps be better if multiple multi criteria methods, along- side multiple decision problems, be employed to derive a reference model for the effectiveness of multi criteria models.


Table Different Techniques and Methods

This paper proposes a goal programming approach to the

selection of software architecture style. This approach can simultaneously handle the multiple and conflicting goals characteristic of the decision problems such as quality, li- mited availability of resources. The integrated AHP-GP- Visualization model applied in three subsequent stages: the first part of the analysis provided the priority levels for the different alternatives (pipe&filter, layered , blackboard) with respect to the criteria (modifiability, scalability, performance,

Method

Multiple

Criteria

Resource

Feasibility

Optimization

required

cost, effort, portability and ease of use). In the second step, the Goal Programming model equations are formulated for



Ranking [14] Yes No No Scoring [15] Yes No No AHP [16] Yes No No Goal Programming [17] No Yes Yes

the selection of the optimal architecture style based of the goals. In the third step, compliance checking is to be ex- ecuted for every single modification made to source code using the SAVE Tool for quality assurance. The application

of the GP technique combined with AHP methodology

DynamicProgramming[18] No Yes Yes

AHP-GP [This Paper] Yes Yes Yes

According to experts, in selecting a style there is no single decision involved but in the decisions considera- tion may be better or worse but still significant. For ex- ample, a style with a low weight might be selected over a style with a high weight if developers are more familiar with the style which has a lower score. The weight vector obtained using AHP for the above example is (0.622,
0.253, 0.101) [19]. Table 18 shows the comparison among the AHP and AHP-GP approaches.

Table Comparison of AHP and AHP-GP approaches

Resources Used

Planning

proved to be a flexible tool to select the best architecture style.

6 REFERENCES

[1] Shaw M., Clements P.: “The golden age of software architec- ture”, IEEE Softw., 2006

[2] Garlan D.: “Software architecture: a roadmap’’, in Finkelstein

A. (ED.): “The future of software engineering” (ACM Press,

2000)

[3] Dobrica L., Niemela E.: “A survey on software architecture analysis methods”, IEEE Trans. Softw. Eng., 2002

[4] Soung Hie K., Jin Woo L.: "An Optimization usability of infor- mation system project resources using a QFD and ZOGP for reflection customer wants", Korea Advanced Inst of Science & Tech.

Method

and design days

Construction

months

Budgeted

cost (00)

Misc

cost (00)

[5] Jin Woo Lee, Soung Hie Kim: “Using analytic hierarchy process

and goal programming for interdependent information system project selection”, Computers & Operation Research

AHP 24 35 300 42

AHP-GP 24 34* 300 24**

* We will save one month construction time as -

2

** We will use only Misc cost $1800 (<$4200) more than the initial Budgeted cost as d 4 = 18 .

[6] Saaty, T. L., “A Scaling Method for Priorities in Hierarchical structures”, Journal of Mathematical Psychology, 1984

[7] Saaty and Takizawa: "The Theory of Ratio Scale estimation", Management Science Vol 33, Nov 1987

[8] Gloria E., Eugene D.and Guisseppi A.,:"A multiple-criteria framework for evaluation of decision support systems", Omega

IJSER © 2011

Volume 32, Issue 4, August 2004

http://www.ijser.org

International Journal of Scientific & Engineering Research Volume 2, Issue 7, July-2011 7

ISSN 2229-5518

[9] Thomas L. Saaty, Luis Gonzalez Vargas: "Decision making with the analytic network process", Management Science and Opera- tion Research, 2006

[10] K. Delhi Babu, P. Govindarajulu, A. Ramamohana Reddy, A.N.

Aruna Kumari : “An Integrated Approach of AHP-GP and Vi- sualization for Selection of Software Architecture: A Frame- work”, International Conference on Advances in Computer En- gineering, 2010.

[11] Slawomir Duszynski, Jens Knodel, Mikael Lindvall : " SAVE: Software Architecture Visualization and Evaluation”, European Conference on Software Maintenance and Reengineering 2009

IEEE.

[12] Ingu Kim, Shangmun Shin: "Development of a Project Selection Method on Information System Using ANP and Fuzzy Logic", World Academy of Science, Engg. and Tech. 2009

[13] Galster M.,Eberlein A. and Moussavi A.:"Systematic selection of software architecture styles", Published in IET Software 2009.

[14] Martin D.J. Buss : "How to Rank Computer Projects", Harvard

Business Review Article, Jan, 1983.

[15] HC Lucas, JR Moor : "A multiple-criterion scoring approach to information system project selection", INFOR 1976

[16] K Muralidhar, R Santhanam : "Using the analytic hierarchy process for information system project selection" Information & Management, Volume 18, Issue 2, February 1990

[17] R Santhanama, K Muralidhara, M Schniederjans : "A zero-one goal programming approach for information system project se- lection", Omega Volume 17, Issue 6, 1989

[18] G. L. Nemhauser, Z. UllmannDiscrete : "Dynamic Program- ming and Capital Allocation", Management Science, Vol. 15, No. 9, Theory Series May, 1969

[19] K Muralidhar, R Santhanam : "Using the analytic hierarchy

process for information system project selection" Information & Management, Volume 18, Issue 2, February 1990

[20] Yubo Gao An AHP-GP Model for Determining Weights, 3"

World Congress on Intelligent Control and Automation, 2000

IJSER © 2011 http://www.ijser.org