Received March 4, 2022, accepted March 15, 2022, date of publication March 23, 2022, date of current version

March 31, 2022.

Digital Object Identifier 10.1109/ACCESS.2022.3161742

MCDM and Various Prioritization Methods in

AHP for CSS: A Comprehensive Review
JAYADEV GYANI 1, (Senior Member, IEEE), AHSAN AHMED 2, AND MOHD ANUL HAQ 1
1 Department of Computer Science, College of Computer and Information Sciences, Majmaah University, Al-Majmaah 11952, Saudi Arabia
2 Department of Information Technology, College of Computer and Information Sciences, Majmaah University, Al-Majmaah 11952, Saudi Arabia

Corresponding author: Ahsan Ahmed ([email protected])

The work of Jayadev Gyani was supported by the Deanship of Scientific Research at Majmaah University under Project R-2022-86.

ABSTRACT Availability and diversity of cloud service providers (CSPs) had put the users into confusion for
its selection of the appropriate service providers. Some cloud service providers are good at some services
while others are good at offering other services. The selection of an appropriate cloud service is one of
the multi-criteria decision analysis (MCDA) problems that became a critical issue of public concern in
the uncertain cloud industry. Based on multiple criteria, various multi-criteria decision-making (MCDM)
methods can be used for the selection of the best CSPs. Researchers considered MCDM techniques as the
best methodology for deciding cloud rank. The paper presents a set of decision criteria and their sub-criteria
required for evaluating CSPs. The main goal of this paper is to present a review of various MCDM methods
for decision-making. Furthermore, the strengths and weaknesses of various MDCM techniques are discussed
to help the researchers about the current trends in the field of decision making. An overview of MCDM
techniques used for Cloud service selection (CSS) is presented. Several methods used for deriving priority
vectors from a Pairwise Consistency Matrix (PCM) in Analytic Hierarchy Process (AHP) technique, used
in recent years are discussed in this research paper.

INDEX TERMS Cloud service providers (CSPs), multi-criteria decision-making (MCDM), cloud service
selection (CSS), pair-wise comparison matrix (PCM), analytic hierarchy process (AHP).

I. INTRODUCTION as a Service, c) IAAS – Infrastructure as a Service. Soft-

In the last few years, due to the evolution of Information and ware as a Service (SAAS) provides on-demand availability
Communication Technology (ICT), Cloud Computing is in of different software as a service to its users through a web
great demand as this technology offers dynamically highly browser. Google App. is an example of SAAS services. With
scalable resources. Cloud Computing has many advantages Platform as a service (PAAS), the cloud provider delivers a
like 24 × 7 availability from any location, reduced cost, fast complete solution for software development. Like SAAS, it
accessibility, better performance, retained standards, compat- also includes Operating Systems and the services required
ibility, effective and efficient management of resources. It is for a particular application [2]. Bungee Connect is a good
an internet-based emerging technology that reduced the cost example of PAAS that provides a complete set of software
of data storage, data processing, and hardware infrastructure development life cycle management tools. Infrastructure as a
as well as software tools. Computing resources collection Service (IAAS) refers to computing and storage infrastruc-
and its management are done automatically using software in ture as a service. It includes servers, networking technology,
cloud computing [1]. The birth of Cloud Computing is origi- and reserved bandwidth for storage. Amazon’s Elastic Com-
nated from Service-Oriented Architecture (SOA). We can say pute Cloud (EC2) is the best example of IAAS that delivers
Cloud Computing is a set of services that are provided from highly scalable computing power. All three services are typ-
the Cloud service providers (CSPs) to different users on its ically charged on a user’s usage basis. It reduces the user’s
demand. The cloud architecture has three different types of stress of buying and setting up the data center for their orga-
layers, a) SAAS – Software as a Service, b) PAAS – Platform nizations. In addition, to the three-layered architecture, the
Cloud deployment models are of various types, namely, Pub-
The associate editor coordinating the review of this manuscript and lic Cloud, Private Cloud, Community Cloud, Virtual Private
approving it for publication was Xujie Li . Cloud, and Hybrid Cloud as shown in Figure 1. Public Cloud

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
33492 VOLUME 10, 2022
J. Gyani et al.: MCDM and Various Prioritization Methods in AHP for CSS: A Comprehensive Review

of this research paper is to review various MCDM methods

available for decision-making. The pros and cons of MCDM
methods are elaborated for the researchers working in the
field of decision making. An overview of MCDM techniques
used for CSS is presented. Various methods used for deriving
priority vectors from Pairwise Consistency Matrix (PCM) in
Analytic Hierarchy Process (AHP) technique, used in recent
years are discussed in this study. This paper is divided into
seven sections as follows: Section I gives the Introduction
about Cloud Computing, the Cloud computing model, and
decision making; Section II presents various decision-criteria
and their sub-criteria for the evaluation of cloud services.
Section III discusses various MCDM methods for decision-
making. Table 2 presents the advantages and disadvan-
tages/limitations of various MDCM techniques. Section IV
presents various MCDM techniques used for CSS. Also, the
FIGURE 1. Cloud Architecture. MCDM techniques used for CSS are also tabulated in Table 3.
Various methods that are used for deriving priority vectors
is set up outside the enterprise firewall and the user is allowed from PCM in AHP are discussed in Section V along with
to access the cloud resources via available API. Security is a a short description given in Table 4. Section VI provides
major issue that critic the usage of public cloud services. The results of this study. Section VII highlights future work and
applications deployed over public clouds need high security conclusion respectively.
from their service providers. Private Cloud refers to setting
up a virtualization environment on its private servers. This II. CRITERIA AND SUB CRITERIA FOR EVALUATING CSPs
type of structure is adopted in those organizations that need Cloud Computing has revolutionized every industry by allow-
full control over their data or infrastructure. A high level ing them to adopt cloud services at cheaper rates. Deciding to
of security is achieved in private clouds as the access is utilize the service for a specific cloud provider is a challeng-
provided only to the shareholders of the organization. But ing task for the client stockholders. Customers get confused
again, it needs more capital investment, so it’s not truly a in determining the basis for selecting appropriate CSPs and
type of cloud service as it needs huge IT infrastructure. Also, their services. A careful analysis is required by cloud users
it is difficult for an individual organization to have a team for the proper selection of the best available CSP from a pool
of security experts that can maintain the level of security of CSPs.
and reliability. Community Cloud is adopted by enterprises Based on our literature study, the criteria and their sub-
having similar type of requirements. Being in an environment criteria that are considered as the main parameter for eval-
of public cloud, getting an isolated private cloud is achieved uating the Quality of service (QoS) of cloud providers are
using the concept of Virtual Private Cloud. In Hybrid Cloud, presented in Table 1. This literature review covers six main
the infrastructure resides partly within the internal enterprise criteria for assessing CSPs.
data center and is partly consumed by third-party cloud ven-
dors [3]. In the hybrid cloud model, there is synchronization A. SECURITY
of resources between the private and public cloud. Hybrid Security plays an important role in adopting cloud infrastruc-
cloud can be used for both internal and external purposes ture, many organization institutions would love to switch to
depending on the user’s need. the cloud, but they are mainly concerned about the security
Decision-making is an aspect of life that we encounter models provided by various CSPs in the market today. On the
every day. It becomes an important issue when confronting other hand, all the vendors guarantee that their cloud infras-
a complex decision-making situation. From a large variety tructure is highly secured and covers all aspects of protection
of CSPs, it has become difficult for companies to adopt the from various malware and vulnerabilities, but the end-users
best CSPs as per their needs. Multi-criteria decision-making are not convinced about pushing their critical data to the
(MCDM) offers a group of methods (shown in Table 2) cloud. Security of data and services must be taken into consid-
for deciding the rank efficiently. Multiple domains have eration seriously to attain the confidence of customers. There
employed MCDM in the decision-making process. Based on are six sub-criteria (Access Management, Privacy, Integrity,
multiple criteria, various MCDM methods can be used for Confidentiality, Incidence Reporting, and Physical Access)
selecting the best CSPs. Table 3 shows Yester year’s work that under this security criteria.
makes use of various MCDM techniques for Cloud service
selection (CSS). B. PERFORMANCE
This paper presents multiple decision criteria and their sub- Every cloud has performance issues that mainly stem from
criteria required for evaluating cloud services. The main goal overall cloud service accuracy, network latency, delay in

VOLUME 10, 2022 33493

 J. Gyani et al.: MCDM and Various Prioritization Methods in AHP for CSS: A Comprehensive Review

TABLE 1. Decision criteria and their sub-criteria for evaluating CSPs.

33494 VOLUME 10, 2022

J. Gyani et al.: MCDM and Various Prioritization Methods in AHP for CSS: A Comprehensive Review

application processing. All these factors are the root cause Various MCDM techniques have been used extensively for
of data loss or delay. Therefore, Cloud providers must offer solving decision-making problems in numerous fields of sci-
a tool for performance monitoring that can help in avoiding ence and technology. Most of the MCDM methods require
potential problems. Also, CSPs must have protocols to miti- explicit weightage of the alternatives. Using these weights
gate the issues that arise in real-time. Performance benchmark of alternatives, the ranking and sorting for the problems are
must be followed by CSPs to easily identify QoS issues done. MCDM approach reduces the incidence of biasing for
as they arise. There are seven sub-criteria (Speed, Accu- any specific problem. This section highlights few techniques
racy, Network Latency, Efficiency, Resiliency, Reliability, used for evaluating the problems based on decision making.
and Interoperability) under the performance criteria. In Mohammed et al. [4] recommended a model based
on eight important identified parameters (i.e., cloud certifi-
C. MIGRATION cations, security issues, polic’s reliability, SLA, Cloud per-
Migration refers to the shifting of resources from one technol- formance, etc.) for evaluating cloud services. Optimization
ogy to another environment. Leading organizations around non-linear technique known as cosine maximization (CM)
the world are migrating their applications to the cloud but technique was used to extract priority vectors for selecting
it is a challenging task, especially with the old applications cloud services. The current research study was based on
that do not support cloud environment. Migration of company the information received using a survey from IT experts
resources to a cloud environment may result in security chal- and faculty members. Although, the CM method indicates
lenges, application downtime, slow application, or data acces- consistency for the pairwise comparison matrix and is more
sibility, and sometimes even need extensive troubleshooting efficient than Euclidean distance and other priority calculat-
to solve the migration issues. Five sub-criteria (Scalability, ing methods. But this method does not support an incom-
Elasticity, Exit Provision, Portability, and Continuity) come plete and inaccurate pairwise comparison matrix. Technique
under this criterion. based on QoS selection can be more helpful and accurate
for analyzing cloud services. A. Hussain et al. [5] introduced
D. AVAILABILITY an integrated technique known as Methodology for Optimal
It is a degree of accessibility of the services delivered by ser- Service Selection (MOSS) for CSS. MOSS method enables
vice providers. To achieve a good reputation for availability, decision-makers to choose the best cloud service with con-
the services provider must operate 24×7. The cloud providers sensus considering both Quality of Service (QoS) and Quality
must adhere to the timeline of the reply mentioned in the SLA. of Experience (QoE). The best worst method (BWM) is used
There are three sub-criteria (Uptime, Downtime, and Outage to obtain weights of two criteria (QoS and QoE). The ranks
Frequency) under this criterion. of various CSPs based on QoS and QoE are evaluated using
existing MCDM methods. The final consolidated ranks of
E. COST CSPs are obtained by Copeland’ method.
Cost is an important factor that causes the consumer to adopt A Trust entity is considered as a type of cloud service that
a particular cloud service. A service consumer always likes helps in increasing the transaction rate in a cloud environ-
to avail the services at a cheaper price. The amount paid ment. This trust-based mechanism constructs some security
by a user to CSPs for using a service is not the total final strategies to safeguard its users. The stakeholders of the
payment. Some additional costs are to be paid to get services cloud, i.e., customers can trust different cloud suppliers, and
ready to use in a business process. Five sub-criteria (Storage similarly, the suppliers can also trust their customers [6]. The
Cost, Processing Cost, Network Cost, Data Transfer Cost, and trust-based model in this paper is based on some attributes
Possession Cost) come under this criterion. like domain name, trust degree, service type, etc. The rep-
utation of cloud customers is completely based on these
F. ACCOUNTABILITY attribute’ values. The cloud providers completely rely on
It refers to the responsibility to use and protect the infor- the parameters like domain name, trust value and generation
mation beyond mere legal requirements. Cloud consumers time, etc. Wrong/fraudulent values of these parameters can
expect privacy and security of their data from CSPs. Addi- give a false reputation for the cloud stack holders and hence
tionally, the data flow is expected to be global and dynamic. cannot be considered as a reliable method for adopting the
It is not only applied to CSPs, both service providers and cus- cloud services in terms of security. Regarding cloud service
tomers are equally involved in this. If any of the parties is not trustworthiness, Sheikh in 2013, introduced a framework for
following the policy rules, it may ruin the company’s reputa- verifying the security controls and capabilities claimed by
tion or may get huge fines. There are four sub-criteria (Com- the cloud providers. The security controls published by the
pliance, Possession, Acceptability, and Reputation) under this providers are expressed as trust properties that are validated
criterion. by some trustable authorities. The proposed hybrid model is
a mixture of hard and soft trust. Trust properties are validated
III. MCDM METHODS FOR DECISION MAKING through digital certificates to measure the level of hard trust.
MCDM is a sub-discipline of operations research that is The former experience and business behavior of the entity
concerned with structuring and solving decision problems. are used for assessing the soft trust [7]. The cloud providers

VOLUME 10, 2022 33495

 J. Gyani et al.: MCDM and Various Prioritization Methods in AHP for CSS: A Comprehensive Review

can cheat the system by flooding fake values that result in the the COMET approach was extended to solve Multi-criteria
wrong reputation and increase the acceptance of unsecured Group Decision Making (MCDGM) problems in a hesi-
cloud services by their customers. This can be avoided by tant fuzzy environment. L-R-type Generalized Fuzzy Num-
eliminating such parameters. bers (GFNs) were used to get the degree of hesitancy for an
M. A. Alsalem et al. [8], made a comprehensive review alternative under a certain criterion. This method provides
for MADM approaches to assist different applications dur- decision-making that is resistant to the phenomenon of rank
ing COVID-19. Several issues and challenges were ana- reversal. S. Faizi et al. [83], proposed a combined approach
lyzed and discussed for multi-attribute, inconsistency, time of Intuitionistic Fuzzy Sets (IFS) and COMET for solving
consumption, unnatural comparison, vagueness, normaliza- MCGDM problems. When compared to HFS, the uncertainty
tion, distance measurement, outranking, trade-off, conflict can be expressed by IFS more accurately. The Triangular
criteria, the importance of criteria, and data variation. Intuitionistic Fuzzy Numbers (TFINs) can be used to handle
Smarandache [73], presents an alternative approach of AHP uncertain data. The methodology requires an adaption of the
called α-Discounting method for MCDM in short α-D matrix of expert judgment (MEJ). The good consistency level
MCDM. A set of preferences are transformed into a system of MEJ, produces a good solution for MCDGM problems.
of linear/non-linear, homogeneous /non-homogeneous and The decision from experts may have a margin of error. The
equality/inequality. AHP works only for preferences that are degree of membership can be represented in terms of inter-
represented by PCM. On the other hand, α-D MCDM can vals and not in crisp numbers. To solve this problem, the
be applied to any number of preferences [74]. As the name COMET, and Normalized Interval-Valued Triangular Fuzzy
implies discount, this method discounts the coefficient of an Numbers (NIVTFNs) can be combined to generate a precise
inconsistent problem to some percentage so that the problem solution in an uncertain environment [84].
can be transformed into a consistent problem. Three examples Jean [85] proposed the Stable Preference Ordering
of α-D MCDM method were presented [75] for the solution Towards Ideal Solution (SPOTIS) method for MCDM based
of non-linear decision-making problems. In [76], the α-D problems. In comparison to COMET, the approach is easy,
MCDM method was used for intervals as a preference instead needs very less information and has low complexity, and is
of crisp numbers. Two consistent and one inconsistent exam- resistant to the phenomenon of rank reversal, as the ranking is
ple were constructed for solving decision-making problems recognized based on the MCDM problems matrix score. It is
and finally, more complicated results were returned. based on the computation of the normalized distance of each
In the field of decision making, the popular approach alternative with respect to the best solution chosen for each
for obtaining a final ranking solution is based on distance. criterion, and their weighted average distance. Nolberto [86],
TOPSIS, VIKOR, CODAS, etc., are some of the techniques demonstrates a sequential interactive model for urban sus-
that use a distance-based approach for evaluating rank. In the tainability (SIMUS) that was based on Linear Programming
case of the TOPSIS technique, it ranks the alternatives with and is resistant to the phenomenon of rank reversal. The
two reference points which sometimes are often insufficient phenomenon of rank reversal was examined by considering
in case of non-linear problems. Additionally, it creates a a situation of two or more projects that have identical values.
problem of rank reversals. In [77], Wojciech, proposed a new This new concept was proved by analyzing the algorithm
distance based MCDM method known as the characteris- used to solve the decision-making problem. In [87], AHP
tic objects method (COMET). COMET is an intuitionistic and SIMUS methods were discussed that uses subjective
approach that uses more reference points and does not require weighting for ranking the alternatives. SIMUS relies on the
weighting factors. The preferences of each alternative are elicitation of experts’ opinions in an objective ranking proce-
achieved based on the distance from the nearest characteristic dure and is based on linear programming (LP). The methods
objects and their values are obtained by using the tournament were used to rank renewable energy sources (RESs) projects
method and the principle of indifference. A fuzzy model is and were proved to be more effective in facilitating MCDGM
being constructed that yields the preference values of the in transparent procedures by enabling communities to make
alternatives, making it a multi-criteria model free of the rank use of their initiative.
reversal phenomenon [78]. In case of uncertainty, the level of reliability in MCDM can
A new variant of the COMET method based on hesi- be increased by reducing the subjectivity and increasing the
tant fuzzy sets (HFS) was proposed by Wojciech [80]. The reality of the obtained results. In [88], Sveta, proposed a fuzzy
approach solves the problem of experts to determine unam- SIMUS approach that is based on the Fuzzy LP method and
biguous membership value for attributes. A case study for the the SIMUS method. Without using the weights, the approach
selection of electric city buses was presented. HFS COMET yields optimal results. Fuzzy SIMUS uses three stages. In the
is resistant to the phenomenon of rank reversal and pro- case of uncertainty, the formation of the parameters of a
duces more reliable decisions by aggregating the uncertain multi-criteria model is done in the first stage. The Fuzzy LP
data. For solving uncertainty problems, in [81], the COMET method is used to form the fuzzy SIMUS model for each
approach was extended using HFS theory. A membership objective in its second stage. The final stage is used to rank the
degree is established as a set of values that helps in facili- alternatives. This methodology was employed in the planning
tating a correct decision for uncertain data. Similarly, in [82], of railway intercity passenger transport in Bulgarian’s railway

33496 VOLUME 10, 2022

J. Gyani et al.: MCDM and Various Prioritization Methods in AHP for CSS: A Comprehensive Review

network. Nine alternative transport plans and eight criteria that impact the trustworthiness of cloud services. Based on
were studied. Verification of the outcomes was accomplished these size factors an attribute model is proposed for the trust-
successfully as the stability of the choice offered a suitable worthiness of cloud services. A method based on the Infor-
alternative. mation Entropy and Markov chain was proposed to measure
Assigning weights to respondents makes the records the degree of uncertainty of each factor used in the proposed
more comprehensible as closely as possible. In this regard, attribute model and measure the level of trustworthiness of
Ping Zhou [9], presents a reliability model to measure the cloud services. To promote the standards of credibility and
reliability of cloud services based on hierarchy variable to increase the level of trustworthiness more factors can be
weights and statistical classification. The model efficiently explained and added to the existing trustworthy attribute
evaluates the reliability of cloud service by hierarchical divi- model. Almost, all cloud services evaluation methods assume
sion of four main characteristics and their sub-characteristic that the various selection criteria used for evaluating cloud
respectively. The research was done in continuation of the services are independent of each other. However, these cri-
earlier work done by P. Zhou et al. [10]. In his old research, teria have some interactions with each other that show an
a quality model for evaluating cloud services was proposed impact on the performance of the selection of cloud services.
based on six main characteristics. The new research done Le Sun [15] proposes a CSSCI framework for the appropriate
by the author is based on four main characteristics. Two selection of the best cloud services. Based on various interac-
characteristics: Completeness and Correctness are removed tions criteria, the non-linear preferences of users were mod-
in the new model and few characteristics like Recoverability eled. This technique can be used to select cloud services based
and Data backup are added. Availability and Continuity are on the use’s criteria priority order (weights) and the interac-
added in place of Stability. Reliability is removed from the tion types (interaction indices) between these priority criteria.
main characteristics and added as a sub characteristic in a new The non-linear constraint optimization technique is used to
model for evaluating cloud reliability more comprehensively. evaluate the Shapley significance and to identify the interac-
Exclusion and replacement of the characteristics were done tion indices of priority criteria. The method is more efficient
based on the evaluation activity which may deviate from in comparison to other MCDM service selection methods.
reality. In addition, there is no clear weight assigning strategy The main issue with this technique is that it is validated only
for CSPs. for crisp data and does not support the preferences of users
L. Coppolino et al. [11], states that multiple factors must and QoS performance for real data. Based on the concept
be taken into consideration by a company for performing of Neutrosophic AHP, Mohamed Abdel-Baset [16], devel-
technology selection. A methodology based on the fuzzy oped the Neutrosophic MCDM methodology for selecting the
logic approach is proposed to evaluate cloud offering selec- best cloud services. Incompatible and ambiguous informa-
tion for a group of companies that have a common supply tion that exists during the process of performance analysis,
chain. A challenging real case study is presented for the is handled by triangular neutrosophic numbers symbolized
manufacturing domain of agro-based companies. To obtain by linguistic variables. The newly induced bias matrix when
the consistency of choices, three triangular fuzzy number sys- used in a neutrosophic environment reflects an improvement
tems are used by the author. Result obtained for the group of in the consistency rate. This method is novel but needs more
companies and the result found for most companies were not involvement of companies for its verification. The method
similar as this approach selects the most suitable solution for also faces challenges to express complex determinate parts.
the entire supply chain without considering the characteristics
of individual companies. In R. R. Kumar [12], [13] introduced IV. VARIOUS MCDM TECHNIQUES USED FOR CSS
a hybrid service selection technique by integrating AHP Cloud computing is transforming ICT industries by offering
weighting methods with TOPSIS (MCSD method). A com- infrastructure, platforms, applications as a service on a sub-
plex problem of cloud services selection is defined by using scription basis. IBM, Microsoft, Google, and Amazon are
AHP. With the help of AHP, criteria weights are computed by the leading enterprises that offer different Cloud services to
pairwise comparison. Final cloud ranking is obtained using their customers. The increase of the computing environment
the TOPSIS method for overall performance. In comparison in every sector leads to an increase in its adoption of various
to other MCDM techniques, this hybrid technique is effective cloud services. In the few past years, considerably many
in evaluating cloud services by providing an accurate result scholars suggested their idea for assessing the services of
based on the requirement of users presented only in quantified CSPs. This section highlights few recent techniques used for
metrics. There is a need for a technique that could evaluate the evaluating the services offered by cloud providers.
cloud services represented by non-quantified metrics. Multiple evaluation criteria are taken into consideration
Before adopting cloud services, a user is always concerned while selecting the optimal CSPs but several decisive factors
about the trustworthiness of service providers. Measuring were neglected. To overcome this problem, Khubaib [62],
the trustworthiness of CSPs is again a key problem. In this introduced a hybrid model using FAHP and WASPAS. A hier-
regard, Z. Ma [14], introduces six factors (Controllability, archical model with 9 main evaluation factors and 30 sec-
cloud service’s visibility, level of user’s satisfaction, viability ondary evaluation factors was prepared. FAHP is used to
of service provider, safety, and reliability of service provider) perform a Relative weight assessment of these main and

VOLUME 10, 2022 33497

 J. Gyani et al.: MCDM and Various Prioritization Methods in AHP for CSS: A Comprehensive Review

TABLE 2. Advantages and Disadvantages/Limitation of various MCDM techniques.

33498 VOLUME 10, 2022

J. Gyani et al.: MCDM and Various Prioritization Methods in AHP for CSS: A Comprehensive Review

TABLE 2. (Continued.) Advantages and Disadvantages/Limitation of various MCDM techniques.

secondary evaluation factors. The rank of the cloud services The system is based on three aspects: The composition of
is evaluated by WASPAS method. The cloud service envi- cloud service that matches with user-requested service(s).
ronment is an uncertain environment that demands methods User preference weights are calculated by the fuzzy AHP
to handle fuzzy information while selecting an appropriate method. Fuzzy TOPSIS is employed to judge cloud services
cloud service. L. Sun [63], presents a novel fuzzy frame- based on the performance and weights of the non-functional
work for improving the existing techniques for CSS. The properties. Considering the objective and subjective attributes
uncertain relationships between the database objects for ser- for judging a cloud computing provider, Sen Lie [64], pro-
vice matching were modeled by building fuzzy ontology. posed a MAGDM based tool that helps client users to select

VOLUME 10, 2022 33499

 J. Gyani et al.: MCDM and Various Prioritization Methods in AHP for CSS: A Comprehensive Review

appropriate cloud providers to fulfill their requirements. the cloud clients for a cloud service, an integrated MCDM
The accurate and efficient result for ranking cloud ven- model was proposed by G. Ilieva et al. [71] that combines
dors is achieved by considering the objective weights of the multi-criteria and fuzzy approaches. The study employed
attributes and experts’ decisions, in addition to the subjective two classic MCDM methods (SAW and WASPAS) to obtain
preferences of the experts and their identity differences. The weight coefficients. Relative weights, crisp values, and lin-
complete procedure involves three steps. In the first prepara- guistics terms were converted into triangular fuzzy numbers.
tion stage, collection of decision opinions is done to prepare Then MARCOS is employed to obtain the ranking of CSPs.
decision matrices, following the normalization process for R. K. Tiwari et al. [72] introduced a framework that identi-
these matrices. The second phase involves the aggregation of fies the best cloud service by using the MCDM method
SV and SAW to determine the attribute weights and deter- TOPSIS and single-valued neutrosophic sets (SVNS). The
mine DM weights by aggregating newly improved TOPSIS framework yields the result by the linguistic rating of cloud
and Delphi - AHP. Aggregate the weights of the attributes services. SVNN (Single-valued neutrosophic number) is an
and preferences of experts by using the LWAA operator. instance of a neutrosophic set that represents uncertain,
In the final stage, the comprehensive evaluation values are imprecise, and incomplete real-world information. SVNN
calculated to decide the best computing vendor. handles inaccurate knowledge for the degree of truth, inde-
H. Ma et al. [65], introduced a time-aware trustworthy terminacy, and falsehood. The newly developed N-TOPSIS
approach for CSS. The theory of interval neutrosophic set method is efficient for the selection of the best cloud ser-
(INS) is adopted to measure three values: performance cost, vice as it is only based on SVNN and not applicable for
potential risk, and level of uncertainty. The cloud service an interval-valued neutrosophic set. Moreover, SVNN and
having high-performance cost, low-performance risk, and MCDM methods other than TOPSIS can be combined to
uncertainty level is considered as the best service. The study improve the consistent result and its efficiency for analyzing
supports decision-making in a cloud environment. The SLA the best CSPs.
between cloud vendors and their clients ensures their com- In [89], Ahmed, used relative preference of various cri-
mitment for QoS provided among them. It is difficult to teria and alternatives to propose an efficient and feasible
evaluate the degree of risk in a virtual environment by the MCDM approach for CSS. It was an integrated approach
traditional QoS web service approach. In [66] based on that makes use of TOPSIS and BWM methods. Weights of
cloud theory, Fan Lin proposed a risk assessment method criteria were acquired by the BWM method, which needs
for cloud computing. Risk values and indicators from virtual less computation as it requires fewer data comparisons and
machines are used to generate five property clouds. Accord- produces reliable results as it does not include second com-
ing to the weight matrix, these five clouds are integrated into parison. TOPSIS method uses weights and relative scores
one cloud with the help of cloud backward generator, thus to evaluate the rank of CSPs. To prove its effectiveness
reducing the problem to determine the risk level of a cloud and reliability against AHP standard method, the integrated
computing environment. A. Jaiswal [67], rank the best cloud approach was tested for use case scenarios by considering
services based on QoS using TOPSIS and fuzzy TOPSIS. nine criteria for the selection of CSPs. Although, the inte-
The weights of multiple criteria yielded by AHP and ANP grated technique was proved an efficient and better approach
respectively were given as an input to TOPSIS and fuzzy to rank CSPs; the use of TOPSIS may lead to the problem
TOPSIS to decide the best cloud service. Due to the capability of rank reversal. Interval-valued intuitionistic fuzzy (IVIF)
of handling unlimited criteria and alternatives [68], Rakesh numbers overcome the weaknesses of conventional crisp
Ranjan Kumar designed a new model by integrating fuzzy numbers by handling uncertain information in real-life appli-
TOPSIS with AHP. Weights obtained by AHP were utilized cations. The IVIF sets are more acceptable to solve complex
by the fuzzy TOPSIS technique for calculating the rank order problems of decision-making [90]. An integrated framework
for CSPs. The sensitivity analysis for the results obtained was provided in [91] to evaluate appropriate CSP. AHP
by this new model shows good robustness for its ranking under IVIF environment was integrated to determine weights
decision and less dependency on criteria weights. To assess of criteria. Other MCDM methods like COPRAS, TOP-
the trustworthiness of CSPs, S. Singh et al. [69], proposed SIS, MULTIMOORA, and VIKOR were integrated under
a CMTE system that makes use of the TOPSIS technique to IVIF environment to assess a set of choices that eventu-
derive trustworthiness from compliance between CSPs and ally analyze the best alternative with vague information.
cloud clients. The system satisfies the cloud user(s) for their The use of MCDM under the IVIF environment is success-
QoS requirement by selecting an appropriate CSP from a pool ful in dealing with ambiguous information. The framework
of CSPs. can be used for solving decision-making complex problems
C. Jatoth et al. [70], proposed an integrated MCDM model with uncertainty. However, the methodology does not deal
for assigning multiple ranks to cloud services based on the with numerous criteria/alternatives and may require complex
quantified QoS characteristics using an extended Grey Tech- calculations and lengthy time. There is a need to investigate
nique for Order of choice along with AHP. This approach this methodology for optimization. Changes in the weights
reduces the uncertainty in data and ambiguity in the pro- of the criteria may also change the result and hence need to
cess of decision-making. Depending on the requirements of observe.

33500 VOLUME 10, 2022

J. Gyani et al.: MCDM and Various Prioritization Methods in AHP for CSS: A Comprehensive Review

A hybrid methodology of BWM and TOPSIS methods was be extended for other fuzzy numbers. In [97], an integrated
proposed by Rakesh [92] to rank the cloud services. It was a MDCM model was proposed for CSS based on balanced
three-phase approach. In the first phase evaluating criteria and scorecard (BSC), fuzzy delphi method (FDM), and FAHP.
alternatives were determined. The second phase was used to Four main criteria were identified as BSC perspectives. FDM
find the weights of these criteria using BWM. The final phase was used to identify decision-making factors under each
used the TOPSIS method to yield the best rank for the services BSC perspective. A decision-making hierarchical model was
offered by CSPs. The method was robust and show good constructed. FAHP was employed to calculate the weights
consistency against the sensitivity analysis. The drawback of for decision-making criteria. Comparison between criteria
this method was its inability to consider interrelationships was done by FAHP to yield the ranking result for the CSS
among the various decision-making criteria. A trust relation- problem. This study considers only fourteen decision-making
ship between cloud consumers and providers revealed the criteria for CSS which is the main limitation for this research
reputation level of CSPs. To evaluate the trustworthiness of work. The numbers of criteria are less and need to consider
CSPs and for the selection of the best CSP, a Context-Aware more criteria for evaluating CSS problems. Additionally, a
Multifaceted Trust Framework (CAMTF) was proposed by sensitivity approach must be performed to validate the pro-
Alhanahnah [93]. CAMTF makes use of two MCDM meth- posed model. Four parameters were analyzed using AHP
ods: AHP was used to compute SLA trust factor while non- for evaluating the rank of the CSPs. The data was provided
SLA trust factor was computed using Fuzzy Simple Additive by a cloud Storage Company called Nasuni. This work was
Weighting (FSAW). It was a versatile system that works implemented for evaluating five CSPs based on four criteria
under diverse conditions. Few issues like automation of SLAs which was a limitation of this research work [98]. A bench-
extraction process were not explored in detail. Also, the sys- marking tool was presented by Piotr [99], which analyzed
tem does not discuss the result for the situation of uncertainty. five criteria using the AHP method to support inexperienced
In the process of pairwise comparison, the classical AHP users for taking correct decisions while choosing the best
ca’t deal with imprecision and subjectivity. To handle this solution from five available CSPs. The study was done for
problem, Sehra et al. proposed a Fuzzy Analytic Hierarchy a country with limited CSPs and limited criteria. However,
Process (FAHP) which deals with uncertain values given smaller companies using few resources like email, ftp may
by the decision-makers. FAHP was used to select the best be benefited from this proposed strategy in making decisions
model for estimating the effort for a given problem. The in the absence of any experts.
work included the comparison of AHP and Fuzzy AHP, and A hybrid MDCM method for CSS was proposed by
a case study was presented to select an effort estimation Al-Faifi [100]. It considers the inter-relationship between the
model [94]. Other MCDM methods can also be experimented performance criteria. K-means algorithm was used to cluster
with the fuzzy approach. A hybrid approach of three fuzzy the CSPs with similar features. The process of clustering
techniques (Fuzzy ANP, Fuzzy TOPSIS, and Fuzzy ELEC- various CSPs reduces the number of matrices to be eval-
TRE) was proposed by Subramanian [95] to solve complex uated by DEMATEL and ANP methods Furthermore, one
decision-making problems involving qualitative and quanti- representative is obtained from each cluster using DEMATEL
tative criteria.Fuzzy set theory handles uncertainty in CSS and ANP techniques. ANP method calculates the weights
in an efficient way. The approach involves three phases. of criteria and finally, the best alternative is obtained. The
Phase one calculates the relative criteria weights using Fuzzy DEMATEL method requires the inter-relationship between
ANP (FANP). In phase two, these weights were used by the alternatives, and this can be the limitation of this hybrid
Fuzzy TOPSIS to produce a weighted normalized matrix that model. Also, the selection of CSPs was based on four criteria.
eventually calculates the rank of the alternatives. The final More criteria must be taken into consideration for selecting
rank of the alternatives were achieved in phase three which the best CSPs. Two processes (FAHP and PROMETHEE)
apply Fuzzy ELECTRE to the three top-ranked results pro- approach were proposed by Boutkhoum [101]. At first, FAHP
duced by phase two. A real case study was done to evaluate was used to structure the criteria and convert the decision of
the proposed approach. Criteria weights were interchanged experts to an appropriate value. Secondly, using the precise
to perform sensitivity analysis of the proposed model. The weights of the alternatives, the PROMETHEE method was
successful result of the sensitivity analysis was a proof of the used to order the alternatives. Results investigated using sen-
robustness of the proposed approach. This approach produces sitivity analysis prove the combined technique as a suitable
the rank by considering uncertain values but was only limited tool for evaluating CSPs but consider a narrow focus on only
to triangular fuzzy values. five cloud computing solutions.
More alike Subramanian, an approach based on Fuzzy
AHP was discussed by Kumar [96]. Quantitative and qual- V. VARIOUS PRIORITIZATION METHODS IN AHP
itative evaluating criteria were identified and PCM was con- Analytic Hierarchy Process (AHP) is the oldest and most
structed using triangular fuzzy numbers. The criteria were extensively used technique for decision-making based on
compared, and the final rank was produced and evaluated the MCDA approach. Based on common characteristics, the
using a case study. Again, this approach also ranks the alter- decision problems are decomposed into elements. Further,
natives using triangular fuzzy values only. The approach can these elements are represented in a hierarchical model with

VOLUME 10, 2022 33501

 J. Gyani et al.: MCDM and Various Prioritization Methods in AHP for CSS: A Comprehensive Review

TABLE 3. Various MCDM techniques used for CSS.

33502 VOLUME 10, 2022

J. Gyani et al.: MCDM and Various Prioritization Methods in AHP for CSS: A Comprehensive Review

TABLE 3. (Continued.) Various MCDM techniques used for CSS.

different levels. Levels can be further subdivided into sub- AHP makes use of priority vector, ω = (ω1 , ω2 , ω3 , . . . , ωn ),
levels depending on the requirement. Comparing the elements derived from the Pairwise Comparison Matrix (PCM). The
of a level with a specific element of its upper level, a pair- various methods used for deriving priority vectors from a
wise comparison matrix (PCM) is prepared by receiving the PCM in AHP are as follows:
rating (scale 1-9) from experts. Generating PCM from the
judgments of experts is one of the difficult tasks in the AHP A. EIGENVECTOR (EV)
technique. Diagonal elements of PCM are equal to 1. As per Priority vector ω of a square matrix A is achieved by solving
Saaty’s definition [42], a PCM is said to be consistent if all a linear problem as [41]
the elements of a matrix hold
aij = aik ∗ akj Aw = λw

VOLUME 10, 2022 33503

 J. Gyani et al.: MCDM and Various Prioritization Methods in AHP for CSS: A Comprehensive Review

Such that ω ni=1 ωi = 1 where ωi ≥ 0 and i =

P
to transform the non-linear optimization problem into a linear
1, 2, 3, . . . , n. equation by performing the differentiation of the DLS method
If λ is an eigenvalue for A then the solution obtained will equation.
be nonzero always. If A with positive values is perfectly
consistent, if and only if λ = n; otherwise λ > n for E. LOGARITHMIC LEAST SQUARES METHOD (LLS)
inconsistency. The consistency ratio value less than 0.1 for LLS is a variation of the WLS method. Due to its sim-
a comparison matrix indicates its acceptance for consistency. ple method of calculation, the method is extensively used.
The procedure to estimate the vector by using the EV method The LLS method minimizes the objective function andob-
is achieved by consecutive squaring the comparison matrix tained the desired Geometric mean vector, ωi = (ω1 , ω2 ,
and each time normalizing the row sums. Stop the procedure ω3 , . . . , ωn )T , as a multiplicative normalizing constraint such
once the difference between normalized sums in two consec- tha:
utive calculations is smaller than a prescribed value. For small X n X n
deviations of consistent ratios ωi /ωj , the EV method produces min (ln aij − ln wi + ln wj )2
the desired output priority vector ω, but this method gives i=1 j=1
Yn
unreliable results if the matrix is of a high order. Such that wi = 1, wi > 0, and i = 1, 2, 3, . . . .n.
i=1
B. ADDITIVE NORMALIZATION (AN) Gordon Crawford in 1985 [45], proved that the formulated
It is one of the simplest methods used to obtain the approx- optimization problem’s solution is unique and can be calcu-
imate priorities vector. Priority vector w is derived by divid- lated as the geometric mean of the elements in each row of
ing the elements of each column of matrix A by the sums square matrix A:
Yn 1
of columns in the comparison matrix. The resulting vector wi = (aij ) n , i = 1, 2, 3, . . . .n.
is obtained by dividing the summation of the elements in j=1
each row by the number of elements in the respective given
F. GOAL PROGRAMMING METHOD (GP)
row [42], [43]. The procedure is as follows:
GP method for producing the priority vector makes use of
aij
a0ij = Pn , where i, j = 1, 2, 3, . . . .n. logarithms of the elements of matrix A [47]. δij+ and δij− are
aij real numbers that are greater than or equal to 1, but both
Pi=1
n 0
j=1 aij cannot be greater than 1. It means:
wi = , where i = 1, 2, 3, . . . .n.
n δij+ = δij− = 1
The resultant priority vector wi obtained by the AN method is
approximately like the solution extracted by the EV method. The priority vector is obtained by minimizing the linear goal
programming problem as follows:
Xn Xn  
C. DIRECT LEAST-SQUARES METHOD (DLS) log2 = minimize log δij+ + log δij−
i=1 j<1
Priority evaluation problem can be expressed into a con-
strained non-linear problem and can be solved by DLS such that
method which is based on optimization approach. The main log vi − log vj + log δij+ − log δ −
ij = logaij , i,
objective of the DLS method is to minimize the Euclidian
j = 1, 2, 3, . . . . . . .n., j > i,
distance. The formulated non-linear optimization problem
has an unlimited solution and hence it disables any random v is an unnormalized vector that will be normalized to produce
selection of priorities vectors. Also, the DLS method does not priority vector w.
preserve rank as well [43], [44].
n G. LOGARITHMIC GOAL PROGRAMMING (LGP)
Xn X
min (aij − wi /wj )2 Mostly all the AHP methods are used for generating priority
i=1
j=1 vectors from PCM, but the LGP method generates a consen-
Xn sus priority vector so that ratio of wi and wj and the value of
Such that wi = 1, wi > 0, and i = 1, 2, 3, . . . .n. at ij specified by decision-makers are almost the same [48].
i=1
pt ij and qt ij are real numbers which are greater than or equal
D. WEIGHTED LEAST SQUARES (WLS) to 1, but both cannot be greater than 1. Subject to
It is an optimization method that eliminates the shortcomings !
pt ij
 
of the DLS method [44], by modifying the DLS objective wi
∗ = at ij
function such that: wj qt ij
n X n
X If pt ij = qt ij = 1, it means that the value stipulated
min (wi − aij wi )2
by decision-makers are consistent otherwise, the values are
i=1 j=1
inconsistent and need a minimization of the product
Solving a non-linear optimization problem numerically is Y Y
quite a tough task. Therefore, the WLS method can be used pt ij qt ij
iI jI

33504 VOLUME 10, 2022

J. Gyani et al.: MCDM and Various Prioritization Methods in AHP for CSS: A Comprehensive Review

If T is the index set and M = |T| then for the entire set of up to acceptable values [53]. The SVD of matrix A can be
PCM values, we need minimization of the product represented by:
Y Y Y
pt ij qt ij A = UDVT ,
tT iI jI

The linear goal programming problem will be solved as

where A is a (m x n) matrix, D is a diagonal (k x k) matrix
follows:
X having positive diagonal elements as α1 , α2 , . . . . . . αk , U and
log2 = minimize (1/M ) lg2t , V are the matrix whose columns are orthonormal. SVD of
t
matrix A in terms of dyads can be written as:
such that
log vi − log vj + log pt ij − log qt ij Xk
A= αi ui vi T ,
= log a ij , i,j = 1, 2, 3, . . . . . . .n., tT ,
t i=1

v is an unnormalized vector that will be normalized to produce Diagonal elements of the diagonal matrix D are known as
normalized consensus priority vector w. singular values, u and v are left and right singular vectors that
form the orthonormal basis for columns and rows of matrix
H. FUZZY PREFERENCE PROGRAMMING (FPP) A in (m x n) dimensional spaces.
Frobenius norm of a matrix is
Mikhailov [50], [51] developed an FPP method to generate
crisp priorities from inconsistent interval or fuzzy comparison r
Xn Xn 
judgments rather than the numerical values. FPP transformed ||A||2 F = a2 ij
i=1 j=1
the prioritization problem into a linear problem.
If interval judgments aij = (lij , uij ) are consistent then Pk∗
i=1 αi ui vi , is a (m x n) matrix with rank k .
A[K∗] = T ∗
component ratios of the priority vectors will satisfy the ∗
It is formed by the largest k singular value of a matrix A
inequalities: and its corresponding singular weight vector. A[K∗] is the
wi rank k∗ least-squares approximation of A that minimizes the
lij ≤ ≤ uij , where lij and uij are the intervals,
wj Frobenius norm to
i = 1, 2, 3, . . . . . . n−1, j = 1, 2, 3, . . . .n,j > i ||A − X ||2F = ni=1 nj=1 (aij − xij )2 for all matrices X of
P P
the rank k∗ or less.
Then
aij wj - wi = 0 and can be represented as a linear equation
J. INTERVAL PRIORITY (IP)
m = n(n−1)2 and matrix form as Rw = 0
Due to the uncertainty of judgments from DM, the weights
And if interval judgments aij = (lij , uij ) are inconsistent,
of priority should be obtained as an interval. In 2004 [55],
it means that the interval judgments are not satisfied by any
derive the estimated weight interval of priorities from PCM.
of the priority vectors. In this scenario, we need to find a
The degree of inconsistency in data is the summation of the
priority vector that can approximately satisfy all the interval
width of the obtained interval priorities.
judgments. Matrix form must be satisfied as
The interval weights Wi is obtained from PCM aij based on
Rw ≈ 0 the following conditions:
First, the PCM aij must be present in the expected interval
The method generates a priority vector and transform the pri- comparison Wij . It means:
oritization problem into a linear program by using FPP. max
µ (µ is the measure of intersection that is also a Consistency wi wi
index-CI) aij ∈ Wij ↔ ≤ aij ≤ , Where wi , wj and wi , wj
wj wj
such that
are upper and lower bound of Wi and Wj respectively.
µdj+ + Rj w ≤ dj+ ,
Second, interval weights Wi will be normalized only if
µdj− − Rj w ≤ dj− , j = 1, 2, 3, . . . .m, 1 ≥ µ ≥ 0;  
Xn X
wi = 1,wi > 0, i = 1, 2, 3, . . . .n. wi − max wj − wj ≥ 1,
i=1 i j
X  
+
dj−and d are tolerance parameters whose values can be set wi − max wj − wj ≤ 1
i j
j
equal for practical implementation of FPP.
Third, the objective function to narrow the estimated interval
I. SINGLE VALUE DECOMPOSITION (SVD) weights Wij is as follows:
SVD is an approach used for deriving associated weight X

vectors in an easy way in AHP. The weight vector associated min wi − wi
wi ,wi i
with the largest singular value of a PCM produces weights

VOLUME 10, 2022 33505

 J. Gyani et al.: MCDM and Various Prioritization Methods in AHP for CSS: A Comprehensive Review

p
K. LINEAR PROGRAMMING METHOD (LP)
Xn Xn aij − āj
= qP
Bala in 2005 [56], developed an approach for producing j=1 i=1 n 2
k=1 (akj − āj )
priority vector. The approach is composed of two stages.
wi − w
A Linear program to establish a consistency bound for a given × qP
n
k=1 (wk − w)2
PCM is formulated in stage one. Further, in the second phase,
the established consistency bound is used in Linear Program
to yield an optimal priority vector. If
Establishing the consistency bound for a given PCM is aij − aj 1 Xn
formulated as:
bij = q
Pn
2 , where aj =
n i=1
aij ,
k=1 akj − aj
Xn−1 Xn
Minimize zij i,j = 1, 2, 3, . . . . . . n and
i=1 j=i+1
wi − w 1 Xn 1
Such that xi − xj − yij = ln aij , ŵi = qP , where w = wi =
n n i=1 n
k=1 (wk − w)
2
i,j = 1, 2, 3, . . . . . . .n. and i 6 = j
zij ≥ yij , i,j = 1, 2, 3, . . . . . . .n. and i < j CCM method can estimate the priorities for a consistent
zij ≥ yji , i,j = 1, 2, 3, . . . . . . .n. and i < j matrix A as follows:
Xn Xn Xn Xn
Here xi = ln (wi ), yij = ln εij , zij = yij . εij is the error that

Max R = bij ŵi = ( bij )ŵi ;
j=1 i=1 i=1 j=1
occurs in the estimation of aij . Xn Xn
such that ŵ2 = 1, ŵi = 0
CI can be evaluated as: i=1 i i=1P
n
j=1 bij
CILP = 2z∗/n(n − 1), Transformed weights ŵ∗i = q 2
,
Pn Pn
i=1 ( j=1 bij )
CILP is the average value of the decision variable zij for
the elements in the above diagonal of the matrix. The first i = 1, 2, 3, . . . . . . n
stage linear program produces a set of all priorities, and it Maximized sum of correlation coefficient
may be possible that multiple solutions exist in this first r
Xn Xn 2
stage. Therefore, to generate the optimal priority vector using R∗ = bij
i=1 j=1
Linear Program, a further minimization of the maximum of
errors will be done as follows: Weight assignment coefficient
Pn Pn
aij − 1 ŵ∗ i − aij ŵ∗ j



Minimize zmax i=1 j=1
β∗ =
2
n ni=1 nj=1 ŵ∗ i − aij ŵ∗ j
P P
Such that
Xn−1 Xn
zij = z∗ , 1
i=1 j=i+1 FinalPriorities w∗ i = + β ŵ∗i ,
n
xi − xj − yij = ln aij , i = 1, 2, 3, . . . . . . n.
i,j = 1, 2, 3, . . . . . . .n. and i 6 = j
zij ≥ yij , i,j = 1, 2, 3, . . . . . . .n. and i < j M. COSINE MAXIMATION (CM)
zij ≥ yji , i,j = 1, 2, 3, . . . . . . .n. and i < j Gang, in 2014 [59], proposed a cosine maximation (CM)
method which increases the cosine similarity measure (CSM)
zmax ≥ zij , i,j = 1, 2, 3, . . . . . . .n.andi < j or the sum of the cosine of an angle between priority vector
zmax is the maximum errors value. and each column vector of a consistency matrix to derive a
reliable priority vector.
L. CORRELATION COEFFICIENT MAXIMIZATION (CCM) In 2019, Mohammed [4] has implemented the CM method
to extract priority vectors for evaluating cloud services. The
Saaty’s definition [42], [58], conclude that a PCM will be
model was based on eight important identified parameters
perfectly consistent if all the elements of a matrix hold
related to cloud certifications, security issues, policies, relia-
A = aij = aik ∗ akj bility, performance, etc. The PCM derived from
Pn Pn

CCM approach for the estimation of priorities from a PCM j=1 i=1 i aij
Maximize C = ω qP qP
was proposed by Ying-Ming [57]. The priorities which are n
ω 2 n 2
k=1 k k=1 akj
highly correlated to each column of a PCM are not consistent
and can be concluded in other ways. As the names imply, Suck that priority vector condition
this approach maximizes the correlation coefficient between  n
X
priorities and each column of a PCM as follows: ω = 1,


i
Xn i=1
Max R = Rj

ωi ≥ 0, i= 1, 2, 3, . . .n

j=1

33506 VOLUME 10, 2022

J. Gyani et al.: MCDM and Various Prioritization Methods in AHP for CSS: A Comprehensive Review

TABLE 4. Prioritization methods in AHP.

Optimal objective function value C∗ is calculated as: in AHP. Free from statistical modeling, the CMM method
r
Xn Xn 2 modifies the entries of the PCM until the Cosine Consistency
∗ Index (CCI) value is not achieved up to 0.90. The method
C = bij
i=1 j=1 is an extension of CM developed by Gang [41], WAM or
In the process of decision-making, improper consistency WGM form is used to modify the matrix to yield a better
may lead to inconsistent results. A consistency improvement consistency rating of ≈ 0.90. Both WAM and WGM form
method is a major approach that eventually increases the uses almost same average number of iterations to achieve a
ranking reputation for a given priority method. The accuracy CCI value ≥ 0.90. NSGA II and SPEA2 are two evolutionary
and reliability of ranking evaluation can be achieved if PCM algorithms [61] that jointly can be used to improve the con-
in AHP is consistent. Adjustment to the comparison matrix sistency index. NSGA II is a variant of the Genetic Algorithm
is done to yield a revised matrix until and unless it does used to perform non-dominated sorting for parent and child
not achieve the value of CR < 0.1. Multiple approaches are populations. Best non-dominated solutions are maintained in
described by researchers for improving the consistency of a fronts. The last front is used to produce solutions based on
PCM in the AHP method. In 2016, Gaurav [60], describes the strategy of crowded distance. The non-dominated solu-
an efficient CMM method to identify the priority vector tions from past generations are stored in an external archive

VOLUME 10, 2022 33507

 J. Gyani et al.: MCDM and Various Prioritization Methods in AHP for CSS: A Comprehensive Review

VII. CONCLUSION AND FUTURE WORK

The need for best decision-making has increased in CSS
during the last few years. Before offering the solutions to any
problem, the proper diagnosis and the level of its accuracy is
very crucial. Decision yielded with in the given timeframe
ascertains its accuracy. Business Analysts intensively use
mathematical-based statistical models, data-orientated meth-
ods such as data mining, machine learning methods, and
MCDM methods for decision-making. This research work
presents various MCDM techniques that are used for ranking
a set of alternatives. AHP is the oldest and most exten-
sively used technique for decision-making based on MCDM
approach. Various methods used for deriving priority vectors
from a PCM in AHP are discussed along with their advan-
tages and limitations. Furthermore, the strengths and weak-
nesses of various MDCM techniques are discussed to help the
researchers about the current trends in the field of decision
making. Although this paper presents most of the recently
used MCDM techniques, there is a lack of acceptance among
FIGURE 2. Choice of MCDM methods for CSS by various researchers. researchers from their perspective of understanding for these
MCDM methods. It is open for a researcher to choose any
TABLE 5. Choice of MCDM methods for CSS by various researchers.
MCDM method depending on one’s interest. Future work
may focus on providing common trends and consistent pri-
ority improvement methods in AHP for CSS.
ACKNOWLEDGMENT
Jayadev Gyani would like to thank the Deanship of Scientific
Research at Majmaah University for supporting this work
under Project Number R-2022-86.
REFERENCES
[1] S. S. Chauhan, E. S. Pilli, R. C. Joshi, G. Singh, and M. C. Govil,
‘‘Brokering in interconnected cloud computing environments: A survey,’’
J. Parallel Distrib. Comput., vol. 133, pp. 193–209, Nov. 2019.
[2] S. S. George and R. S. Pramila, ‘‘A review of different techniques in cloud
computing,’’ Mater. Today, Proc., vol. 46, pp. 8002–8008, Jan. 2021.
[3] S. Sengupta, V. Kaulgud, and V. S. Sharma, ‘‘Cloud computing security-
trends and research directions,’’ in Proc. IEEE World Congr. Services,
Jul. 2011, pp. 524–531.
[4] M. Alshehri, ‘‘An effective mechanism for selection of a cloud ser-
vice provider using cosine maximization method,’’ Arabian J. Sci. Eng.,
vol. 44, no. 11, pp. 9291–9300, 2019.
[5] A. Hussain, J. Chun, and M. Khan, ‘‘A novel customer-centric method-
ology for optimal service selection (MOSS) in a cloud environment,’’
Future Gener. Comput. Syst., vol. 105, pp. 562–580, Apr. 2020.
[6] W. Li, L. Ping, and X. Pan, ‘‘Use trust management module to achieve
effective security mechanisms in cloud environment,’’ in Proc. Int. Conf.
Electron. Inf. Eng., vol. 1, Aug. 2010, p. V1-14.
maintained by SPEA2. After each generation, the archive [7] S. M. Habib, V. Varadharajan, and M. Mühlhäuser, ‘‘A trust-aware frame-
gets updated and SPEA2 computes a strength value for each work for evaluating security controls of service providers in cloud mar-
ketplaces,’’ in Proc. 12th IEEE Int. Conf. Trust, Secur. Privacy Comput.
solution. Commun., Jul. 2013, pp. 459–468.
[8] M. A. Alsalem, R. Mohammed, O. S. Albahri, A. A. Zaidan,
VI. RESULTS
A. H. Alamoodi, K. Dawood, A. Alnoor, A. S. Albahri, B. B. Zaidan,
The decision for the selection of cloud service is a challenging
U. Aickelin, H. Alsattar, M. Alazab, and F. Jumaah, ‘‘Rise of multiat-
issue. In the literature review, various methods based on tribute decision-making in combating COVID-19: A systematic review
MCDM approach are discussed for the problems of CSS. of the state-of-the-art literature,’’ Int. J. Intell. Syst., pp. 1–11, Oct. 2021.
This work analyzed various contributions of the researchers. [9] P. Zhou, L.-M. Meng, X.-S. Qiu, Z.-S. Wang, Z.-P. Wang, and Z.-F. Chen,
‘‘Evaluation of cloud service reliability based on classified statistics and
22 articles on cloud selection using MCDM methods are hierarchy variable weight,’’ J. Signal Process. Syst., vol. 91, no. 10,
reviewed, and it is concluded that majority of the researchers pp. 1115–1126, Oct. 2019.
consider AHP as the best technique for CSS. The choice of [10] P. Zhou, Z. Wang, W. Li, and N. Jiang, ‘‘Quality model of cloud service,’’
in Proc. IEEE 17th Int. Conf. High Perform. Comput. Commun., IEEE 7th
selection of various MCDM techniques for CSS is shown in Int. Symp. Cyberspace Saf. Secur., IEEE 12th Int. Conf. Embedded Softw.
Table 5 and graphically in Figure 2. Syst., Aug. 2015, pp. 1418–1423.

33508 VOLUME 10, 2022

J. Gyani et al.: MCDM and Various Prioritization Methods in AHP for CSS: A Comprehensive Review

[11] L. Coppolino, L. Romano, A. Scaletti, and L. Sgaglione, ‘‘Fuzzy set [34] P. Chatterjee, V. M. Athawale, and S. Chakraborty, ‘‘Selection of indus-
theory-based comparative evaluation of cloud service offerings: An agro- trial robots using compromise ranking and outranking methods,’’ Robot.
food supply chain case study,’’ Technol. Anal. Strategic Manage., vol. 33, Comput.-Integr. Manuf., vol. 26, no. 5, pp. 483–489, Oct. 2010.
no. 8, pp. 900–913, Aug. 2021. [35] J. R. Figueira, S. Greco, B. Roy, and R. Słowiński, ‘‘An overview of
[12] R. R. Kumar and C. Kumar, ‘‘A multi criteria decision making method ELECTRE methods and their recent extensions,’’ J. Multi-Criteria Decis.
for cloud service selection and ranking,’’ Int. J. Ambient Comput. Intell., Anal., vol. 20, nos. 1–2, pp. 61–85, Jan. 2013.
vol. 9, no. 3, pp. 1–14, Jul. 2018. [36] M. Behzadian, R. B. Kazemzadeh, A. Albadvi, and M. Aghdasi,
[13] R. R. Kumar, S. Mishra, and C. Kumar, ‘‘A novel framework for cloud ‘‘PROMETHEE: A comprehensive literature review on methodologies
service evaluation and selection using hybrid MCDM methods,’’ Arabian and applications,’’ Eur. J. Oper. Res., vol. 200, no. 1, pp. 198–215, 2010.
J. Sci. Eng., vol. 43, no. 12, pp. 7015–7030, Dec. 2018. [37] X. Wu and H. Liao, ‘‘An approach to quality function deployment based
[14] Z. Ma, R. Jiang, M. Yang, T. Li, and Q. Zhang, ‘‘Research on the on probabilistic linguistic term sets and ORESTE method for multi-
measurement and evaluation of trusted cloud service,’’ Soft Comput., expert multi-criteria decision making,’’ Inf. Fusion, vol. 43, pp. 13–26,
vol. 22, no. 4, pp. 1247–1262, Feb. 2018. Sep. 2018.
[15] L. Sun, H. Dong, O. K. Hussain, F. K. Hussain, and A. X. Liu, ‘‘A frame- [38] V. Podvezko, ‘‘The comparative analysis of MCDA methods SAW and
work of cloud service selection with criteria interactions,’’ Future Gener. COPRAS,’’ Eng. Econ., vol. 22, no. 2, pp. 134–146, Apr. 2011.
Comput. Syst., vol. 94, pp. 749–764, May 2019.
[39] A. Podviezko and V. Podvezko, ‘‘Absolute and relative evaluation of
[16] M. Abdel-Basset, M. Mohamed, and V. Chang, ‘‘NMCDA: A framework socio-economic objects based on multiple criteria decision making meth-
for evaluating cloud computing services,’’ Future Gener. Comput. Syst., ods,’’ Eng. Econ., vol. 25, no. 5, pp. 522–529, Dec. 2014.
vol. 86, pp. 12–29, Sep. 2018.
[40] A. S. Yalcin, H. S. Kilic, and D. Delen, ‘‘The use of multi-criteria
[17] W. Ma, X. Luo, and Y. Jiang, ‘‘Multicriteria decision making with cogni-
decision-making methods in business analytics: A comprehensive liter-
tive limitations: A DS/AHP-based approach,’’ Int. J. Intell. Syst., vol. 32,
ature review,’’ Technol. Forecasting Social Change, vol. 174, Jan. 2022,
no. 7, pp. 686–721, Jul. 2017.
Art. no. 121193.
[18] G. I. Alptekin and S. E. Alptekin, ‘‘Design of customer-oriented cloud
[41] T. L. Saaty, ‘‘A scaling method for priorities in hierarchical structures,’’
products,’’ in Proc. World Congr. Eng. Comput. Sci. (WCECS), vol. 2,
J. Math. Psychol., vol. 15, no. 3, pp. 234–281, 1977.
2014, pp. 1–6.
[19] E. Schulze-González, J.-P. Pastor-Ferrando, and P. Aragonés-Beltrán, [42] T. L. Saaty, The Analytic Hierarchy Process. New York, NY,
‘‘Testing a recent DEMATEL-based proposal to simplify the use of ANP,’’ USA: McGrawHill, 1980.
Mathematics, vol. 9, no. 14, p. 1605, Jul. 2021. [43] B. Srdjevic, ‘‘Combining different prioritization methods in the ana-
[20] J. Rezaei, ‘‘Best-worst multi-criteria decision-making method,’’ Omega, lytic hierarchy process synthesis,’’ Comput. Oper. Res., vol. 32, no. 7,
vol. 53, pp. 49–57, Jun. 2015. pp. 1897–1919, Jul. 2005.
[21] S. J. Sadjadi and M. Karimi, ‘‘Best-worst multi-criteria decision-making [44] A. T. W. Chu, R. E. Kalaba, and K. Spingarn, ‘‘A comparison of two meth-
method: A robust approach,’’ Decis. Sci. Lett., vol. 7, no. 4, pp. 323–340, ods for determining the weights of belonging to fuzzy sets,’’ J. Optim.
2018. Theory Appl., vol. 27, no. 4, pp. 531–538, Apr. 1979.
[22] N. Kundakci, ‘‘Combined multi-criteria decision making approach based [45] G. Crawford and C. Williams, ‘‘A note on the analysis of subjective
on macbeth and multi-MOORA methods,’’ Alphanumeric J., vol. 4, no. 1, judgment matrices,’’ J. Math. Psychol., vol. 29, no. 4, pp. 387–405,
pp. 17–26, Apr. 2016. Dec. 1985.
[23] R. Rohmatulloh and S. Winarni, ‘‘TOPSIS method for determining the [46] Z. Xu, ‘‘On consistency of the weighted geometric mean complex judge-
priority of strategic training program,’’ Int. J. Adv. Sci., Eng. Inf. Technol., ment matrix in AHP,’’ Eur. J. Oper. Res., vol. 126, no. 3, pp. 683–687,
vol. 4, no. 2, pp. 67–70, 2014. Nov. 2000.
[24] L. Kraujalienė, ‘‘Comparative analysis of multicriteria decision-making [47] N. Bryson, ‘‘A goal programming method for generating priority vec-
methods evaluating the efficiency of technology transfer,’’ Bus., Manage. tors,’’ J. Oper. Res. Soc., vol. 46, no. 5, pp. 641–648, May 1995.
Educ., vol. 17, pp. 72–93, Aug. 2019. [48] N.-M. Bryson and A. Joseph, ‘‘Generating consensus priority point vec-
[25] Z. P. Tian, J. Q. Wang, and H. Y. Zhang, ‘‘An integrated approach for tors: A logarithmic goal programming approach,’’ Comput. Oper. Res.,
failure mode and effects analysis based on fuzzy best-worst, relative vol. 26, no. 6, pp. 637–643, May 1999.
entropy, and VIKOR methods,’’ Appl. Soft Comput., vol. 72, pp. 636–646, [49] G. Dutta, S. Basu, and J. John, ‘‘Development of utility function for life
Nov. 2018. insurance buyers in the Indian market,’’ J. Oper. Res. Soc., vol. 61, no. 4,
[26] M. Keshavarz Ghorabaee, E. K. Zavadskas, Z. Turskis, and J. Antuchevi- pp. 585–593, Apr. 2010.
ciene, ‘‘A new combinative distance-based assessment (CODAS) method [50] L. Mikhailov, ‘‘A fuzzy programming method for deriving priorities
for multi-criteria decision-making,’’ Econ. Comput. Econ. Cybern. Stud. in the analytic hierarchy process,’’ J. Oper. Res. Soc., vol. 51, no. 3,
Res., vol. 50, no. 3, pp. 25–44, 2016. pp. 341–349, 2000.
[27] X. Peng and J. Dai, ‘‘Research on the assessment of classroom teaching [51] L. Mikhailov, ‘‘Fuzzy analytical approach to partnership selection in
quality with Q-rung orthopair fuzzy information based on multiparamet- formation of virtual enterprises,’’ Omega, vol. 30, no. 5, pp. 393–401,
ric similarity measure and combinative distance-based assessment,’’ Int. Oct. 2002.
J. Intell. Syst., vol. 34, no. 7, pp. 1588–1630, 2019.
[52] Y.-M. Wang and K.-S. Chin, ‘‘Fuzzy analytic hierarchy process: A log-
[28] J.-J. Wang, Z.-H. Miao, F.-B. Cui, and H.-C. Liu, ‘‘Robot evaluation and
arithmic fuzzy preference programming methodology,’’ Int. J. Approx.
selection with entropy-based combination weighting and cloud TODIM
Reasoning, vol. 52, no. 4, pp. 541–553, Jun. 2011.
approach,’’ Entropy, vol. 20, no. 5, p. 349, May 2018.
[53] S. I. Gass and T. Rapcsák, ‘‘Singular value decomposition in AHP,’’ Eur.
[29] E. Mulliner, N. Malys, and V. Maliene, ‘‘Comparative analysis of MCDM
J. Oper. Res., vol. 154, no. 3, pp. 573–584, May 2004.
methods for the assessment of sustainable housing affordability,’’ Omega,
vol. 59, pp. 146–156, Mar. 2016. [54] T. D. Braun, R. Ulrey, A. A. Maciejewski, and H. J. Siegel, ‘‘Parallel
[30] F. Tscheikner-Gratl, P. Egger, W. Rauch, and M. Kleidorfer, ‘‘Comparison approaches for singular value decomposition as applied to robotic manip-
of multi-criteria decision support methods for integrated rehabilitation ulator Jacobians,’’ Int. J. Parallel Program., vol. 30, no. 1, pp. 1–35, 2002.
prioritization,’’ Water, vol. 9, no. 2, p. 68, Jan. 2017. [55] K. Sugihara, H. Ishii, and H. Tanaka, ‘‘Interval priorities in AHP by inter-
[31] B. Kizielewicz and A. Bączkiewicz, ‘‘Comparison of fuzzy TOPSIS, val regression analysis,’’ Eur. J. Oper. Res., vol. 158, no. 3, pp. 745–754,
fuzzy VIKOR, fuzzy WASPAS and fuzzy MMOORA methods in the Nov. 2004.
housing selection problem,’’ Proc. Comput. Sci., vol. 192, pp. 4578–4591, [56] B. Chandran, B. Golden, and E. Wasil, ‘‘Linear programming models
Jan. 2021. for estimating weights in the analytic hierarchy process,’’ Comput. Oper.
[32] E. O. B. Nara, D. C. Sordi, J. L. Schaefer, J. N. C. Schreiber, I. C. Baierle, Res., vol. 32, no. 9, pp. 2235–2254, Sep. 2005.
M. A. Sellitto, and J. C. Furtado, ‘‘Prioritization of OHS key perfor- [57] Y.-M. Wang, C. Parkan, and Y. Luo, ‘‘Priority estimation in the AHP
mance indicators that affecting business competitiveness—A demon- through maximization of correlation coefficient,’’ Appl. Math. Model.,
stration based on MAUT and neural networks,’’ Saf. Sci., vol. 118, vol. 31, no. 12, pp. 2711–2718, Dec. 2007.
pp. 826–834, Oct. 2019. [58] T. L. Saaty, Fundamentals of Decision Making and Priority Theory
[33] M. Velasquez and P. T. Hester, ‘‘An analysis of multi-criteria decision With the Analytic Hierarchy Process, vol. 6. Pittsburgh, PA, USA: RWS
making methods,’’ Int. J. Oper. Res., vol. 10, no. 2, pp. 56–66, 2013. Publications, 2000.

VOLUME 10, 2022 33509

 J. Gyani et al.: MCDM and Various Prioritization Methods in AHP for CSS: A Comprehensive Review

[59] G. Kou and C. Lin, ‘‘A cosine maximization method for the priority vector [81] S. Faizi, T. Rashid, W. Sałabun, S. Zafar, and J. Wątróbski, ‘‘Decision
derivation in AHP,’’ Eur. J. Oper. Res., vol. 235, no. 1, pp. 225–232, making with uncertainty using hesitant fuzzy sets,’’ Int. J. Fuzzy Syst.,
May 2014. vol. 20, no. 1, pp. 93–103, Jan. 2018.
[60] G. Khatwani and A. K. Kar, ‘‘Improving the cosine consistency index [82] S. Faizi, W. Sałabun, T. Rashid, J. Wątróbski, and S. Zafar, ‘‘Group
for the analytic hierarchy process for solving multi-criteria decision decision-making for hesitant fuzzy sets based on characteristic objects
making problems,’’ Appl. Comput. Informat., vol. 13, no. 2, pp. 118–129, method,’’ Symmetry, vol. 9, no. 8, p. 136, Jul. 2017.
Jul. 2017. [83] S. Faizi, W. Sałabun, T. Rashid, S. Zafar, and J. Wątróbski, ‘‘Intuition-
[61] A. Bose, ‘‘Using genetic algorithm to improve consistency and retain istic fuzzy sets in multi-criteria group decision making problems using
authenticity in the analytic hierarchy process,’’ Opsearch, vol. 57, no. 4, the characteristic objects method,’’ Symmetry, vol. 12, no. 9, p. 1382,
pp. 1070–1092, Dec. 2020. Aug. 2020.
[62] K. A. Alam, R. Ahmed, F. S. Butt, S.-G. Kim, and K.-M. Ko, [84] S. Faizi, W. Sałabun, S. Ullah, T. Rashid, and J. Więckowski,
‘‘An uncertainty-aware integrated fuzzy AHP-WASPAS model to eval- ‘‘A new method to support decision-making in an uncertain envi-
uate public cloud computing services,’’ Proc. Comput. Sci., vol. 130, ronment based on normalized interval-valued triangular fuzzy num-
pp. 504–509, Jan. 2018. bers and COMET technique,’’ Symmetry, vol. 12, no. 4, p. 516,
[63] L. Sun, J. Ma, Y. Zhang, H. Dong, and F. K. Hussain, ‘‘Cloud-FuSeR: Apr. 2020.
Fuzzy ontology and MCDM based cloud service selection,’’ Future [85] J. Dezert, A. Tchamova, D. Han, and J.-M. Tacnet, ‘‘The SPOTIS
Gener. Comput. Syst., vol. 57, pp. 42–55, Apr. 2016. rank reversal free method for multi-criteria decision-making support,’’
[64] S. Liu, F. T. S. Chan, and W. Ran, ‘‘Decision making for the selection in Proc. IEEE 23rd Int. Conf. Inf. Fusion (FUSION), Jul. 2020,
of cloud vendor: An improved approach under group decision-making pp. 1–8.
with integrated weights and objective/subjective attributes,’’ Expert Syst. [86] N. Munier, ‘‘A new approach to the rank reversal phenomenon in MCDM
Appl., vol. 55, pp. 37–47, Aug. 2016. with the SIMUS method,’’ Multiple Criteria Decis. Making, vol. 11,
[65] H. Ma, Z. Hu, K. Li, and H. Zhang, ‘‘Toward trustworthy cloud service pp. 137–152, 2016.
selection: A time-aware approach using interval neutrosophic set,’’ J. Par- [87] K. Nigim, N. Munier, and J. Green, ‘‘Pre-feasibility MCDM tools to
allel Distrib. Comput., vol. 96, pp. 75–94, Oct. 2016. aid communities in prioritizing local viable renewable energy sources,’’
[66] F. Lin, W. Zeng, L. Yang, Y. Wang, S. Lin, and J. Zeng, ‘‘Cloud computing Renew. Energy, vol. 29, no. 11, pp. 1775–1791, Sep. 2004.
system risk estimation and service selection approach based on cloud [88] S. Stoilova and N. Munier, ‘‘A novel fuzzy SIMUS multicriteria decision-
focus theory,’’ Neural Comput. Appl., vol. 28, no. 7, pp. 1863–1876, making Method. An application in railway passenger transport planning,’’
Jul. 2017. Symmetry, vol. 13, no. 3, p. 483, Mar. 2021.
[67] A. Jaiswal and R. B. Mishra, ‘‘Cloud service selection using TOPSIS and [89] A. E. Youssef, ‘‘An integrated MCDM approach for cloud ser-
fuzzy TOPSIS with AHP and ANP,’’ in Proc. Int. Conf. Mach. Learn. Soft vice selection based on TOPSIS and BWM,’’ IEEE Access, vol. 8,
Comput., Jan. 2017, pp. 136–142. pp. 71851–71865, 2020.
[68] R. R. Kumar, S. Mishra, and C. Kumar, ‘‘Prioritizing the solution [90] G. Büyüközkan, F. Göçer, and O. Feyzioǧlu, ‘‘Cloud computing technol-
of cloud service selection using integrated MCDM methods under ogy selection based on interval valued intuitionistic fuzzy COPRAS,’’ in
fuzzy environment,’’ J. Supercomput., vol. 73, no. 11, pp. 4652–4682, Advances in Fuzzy Logic and Technology. Cham, Switzerland: Springer,
Nov. 2017. 2017, pp. 318–329.
[91] G. Büyüközkan, F. Göçer, and O. Feyzioğlu, ‘‘Cloud computing
[69] S. Singh and J. Sidhu, ‘‘Compliance-based multi-dimensional trust
technology selection based on interval-valued intuitionistic fuzzy
evaluation system for determining trustworthiness of cloud ser-
MCDM methods,’’ Soft Comput., vol. 22, no. 15, pp. 5091–5114,
vice providers,’’ Future Gener. Comput. Syst., vol. 67, pp. 109–132,
Aug. 2018.
Feb. 2017.
[92] R. R. Kumar, B. Kumari, and C. Kumar, ‘‘CCS-OSSR: A framework
[70] C. Jatoth, G. R. Gangadharan, U. Fiore, and R. Buyya, ‘‘SEL-
based on hybrid MCDM for optimal service selection and ranking of
CLOUD: A hybrid multi-criteria decision-making model for selection
cloud computing services,’’ Cluster Comput., vol. 24, no. 2, pp. 867–883,
of cloud services,’’ Soft Comput., vol. 23, no. 13, pp. 4701–4715,
Jun. 2021.
Jul. 2019.
[93] M. Alhanahnah, P. Bertok, Z. Tari, and S. Alouneh, ‘‘Context-
[71] G. Ilieva, T. Yankova, V. Hadjieva, R. Doneva, and G. Totkov, ‘‘Cloud
aware multifaceted trust framework for evaluating trustworthiness of
service selection as a fuzzy multi-criteria problem,’’ TEM J., vol. 9, no. 2,
cloud providers,’’ Future Gener. Comput. Syst., vol. 79, pp. 488–499,
p. 484, 2020.
Feb. 2018.
[72] R. Kumar Tiwari and R. Kumar, ‘‘A framework for prioritizing cloud [94] S. KaurSehra, Y. S. Brar, and N. Kaur, ‘‘Multi criteria decision making
services in neutrosophic environment,’’ J. King Saud Univ.-Comput. Inf. approach for selecting effort estimation model,’’ Int. J. Comput. Appl.,
Sci., pp. 1–16, May 2020. vol. 39, no. 1, pp. 10–17, Feb. 2012.
[73] F. Smarandache, ‘‘α-discounting method for multi-criteria decision mak- [95] T. Subramanian and N. Savarimuthu, ‘‘Cloud service evaluation and
ing (α-D MCDM),’’ in Proc. Fusion Int. Conf., Edinburgh, Scotland, selection using fuzzy hybrid MCDM approach in marketplace,’’ Int. J.
no. 2, Jul. 2010, pp. 29–42. Fuzzy Syst. Appl., vol. 5, no. 2, pp. 118–153, Apr. 2016.
[74] F. Smarandache, ‘‘α-discounting method for multi-criteria decision mak- [96] R. R. Kumar and C. Kumar, ‘‘An evaluation system for cloud service
ing (α-D MCDM),’’ SCS AdSumus, Oradea, Romania Educ. Publisher, selection using fuzzy AHP,’’ in Proc. 11th Int. Conf. Ind. Inf. Syst. (ICIIS),
Columbus, OH, USA, Tech. Rep. 60, 2015, pp. 1–60. Dec. 2016, pp. 821–826.
[75] F. Smarandache, ‘‘Three non-linear α-discounting MCDM-method exam- [97] S. Lee and K.-K. Seo, ‘‘A hybrid multi-criteria decision-making model
ples,’’ in Proc. Int. Conf. Adv. Mech. Syst., 2013, pp. 174–176. for a cloud service selection problem using BSC, fuzzy delphi method
[76] F. Smarandache, ‘‘Interval α-discounting method for MDCM,’’ in and fuzzy AHP,’’ Wireless Pers. Commun., vol. 86, no. 1, pp. 57–75,
Proc. 24th Annu. Symp. Inst. Solid Mech. Session Commission Acoust. Jan. 2016.
(SISOM), Bucharest, Romania, May 2013, pp. 1–8. [98] R. K. Chahal and S. Singh, ‘‘AHP-based ranking of cloud-service
[77] W. Sałabun, ‘‘The characteristic objects method: A new distance-based providers,’’ in Information Systems Design and Intelligent Applications
approach to multicriteria decision-making problems,’’ J. Multi-Criteria (Advances in Intelligent Systems and Computing), vol. 433. Chandigarh,
Decis. Anal., vol. 22, nos. 1–2, pp. 37–50, Jan. 2015. India: Springer, 2016, pp. 491–499.
[78] W. Sałabun, ‘‘The characteristics objects method: A new approach to [99] P. Niemcewicz, ‘‘The use of the multi-criteria AHP method to select a
identify a multi-criteria group decision-making model,’’ Int. J. Comput. cloud computing provider,’’ Proc. Comput. Sci., vol. 192, pp. 2558–2567,
Technol. Appl., vol. 5, no. 5, pp. 1597–1602, 2014. Jan. 2021.
[79] A. Piegat and W. Sałabun, ‘‘Identification of a multicriteria decision- [100] A. Al-Faifi, B. Song, M. M. Hassan, A. Alamri, and A. Gumaei, ‘‘A hybrid
making model using the characteristic objects method,’’ Appl. Comput. multi criteria decision method for cloud service selection from smart
Intell. Soft Comput., vol. 2014, pp. 1–14, Jan. 2014. data,’’ Future Gener. Comput. Syst., vol. 93, pp. 43–57, Apr. 2019.
[80] W. Salabun, A. Karczmarczyk, and J. Wątróbski, ‘‘Decision-making [101] O. Boutkhoum, M. Hanine, T. Agouti, and A. Tikniouine, ‘‘Selection
using the hesitant fuzzy sets COMET method: An empirical study of the problem of cloud solution for big data accessing: Fuzzy PROMETHEE
electric city buses selection,’’ in Proc. IEEE Symp. Ser. Comput. Intell. as a proposed methodology,’’ J. Digit. Inf. Manage., vol. 14, no. 6,
(SSCI), Nov. 2018, pp. 1485–1492. pp. 368–382, 2016.

33510 VOLUME 10, 2022

J. Gyani et al.: MCDM and Various Prioritization Methods in AHP for CSS: A Comprehensive Review

JAYADEV GYANI (Senior Member, IEEE) MOHD ANUL HAQ received the bachelor’s
received the master’s degree in computer science degree from Hemvati Nandan Bahuguna Garhwal
and engineering from Osmania University, India, University, Srinagar, Uttarakhand, the master’s
in 1994, and the Ph.D. degree in computer science degree in computer applications from UP Tech-
from the University of Hyderabad, India, in 2009. nical University (currently Dr. A. P. J. Abdul
He worked as a Lecturer, an Assistant Professor, Kalam Technical University Uttar Pradesh), and
a Professor, and the Head of the Computer Science the Ph.D. degree from the Indian Institute of Tech-
Department. Since 2015, he has been working with nology Roorkee, India, in 2013. He completed
the Department of Computer Science, College of several research projects sponsored by different
Computer and Information Sciences, Majmaah national/international agencies. He was invited
University, Saudi Arabia. He had a teaching experience of 25 years. His from Microsoft HQ, Redmond, to showcase his AI research projects for
research interests include software engineering, big data analytics, dis- the AI for Earth summit in 2019. His research interests include artificial
tributed computing, machine learning algorithms, and their applications. intelligence and machine learning. The target applications of his research
He is a member of ACM. are deep learning-based image classification, modeling, and forecasting.
He serves as a guest editor and reviewer in reputed journals, including Nature,
Elsevier, Springer, Tech Science Press, and many more.
AHSAN AHMED received the master’s degree in
computer science from Jamia Hamdard University,
New Delhi, India, in 2008. From 2008 to 2010,
he was a Research Intern at the CSIR’s Laboratory.
Since November 2010, he has been working as
a Lecturer with the Department of Information
Technology, College of Computer and Informa-
tion Sciences, Majmaah University, Al-Majmaah,
Saudi Arabia. He had 11 years of experience in
teaching. He has presented four papers in interna-
tional conferences and published nine research papers in reputed journals
indexed by Scopus and SCI under his research and other technical areas.
His research interests include machine learning, web technologies, cloud
computing, reputation systems, and e-learning.

VOLUME 10, 2022 33511