Received March 15, 2019, accepted March 28, 2019, date of publication April 2, 2019, date of current version

May 20, 2019.

Digital Object Identifier 10.1109/ACCESS.2019.2908919

Neutrosophic Multi-Criteria Decision Making

Approach for IoT-Based Enterprises
NADA A. NABEEH 1 , MOHAMED ABDEL-BASSET 2, HAITHAM A. EL-GHAREEB1 ,
AND AHMED ABOELFETOUH1
1 Information Systems Department, Faculty of Computers and Information Sciences, Mansoura University, Mansoura 35516, Egypt
2 Department of Decision Support, Faculty of Computers and Informatics, Zagazig University, Zagazig 44519, Egypt
Corresponding author: Mohamed Abdel-Basset ([email protected])

ABSTRACT The Internet of Things (IoT) connects billions of devices to afford inventive opportunities
between things and people. The rapid development of products related to the IoT is a new challenge to keep
security issues, lack of confidence, and understanding of the IoT. Analytical hierarchy process (AHP) is a
classic multi-criteria decision making (MCDM) method used to analyze and scale complex problems and
to obtain weights for the selected criteria. The vague and inconsistent information in real situations can
lead to the decision maker’s confusion. The decision makers cannot determine accurate judgments for all
situations due to the conditions of uncertainty factors in real life; in addition to the limited knowledge and
experience of decision makers. In this research, we present a neutrosophic AHP of the IoT in enterprises
to help decision makers to estimate the influential factors. The estimation of influential factors can affect
the success of the IoT-related enterprise. This study combines AHP methods with neutrosophic techniques
to effectively present the criteria related to influential factors. The recommended alternatives are presented
based on neutrosophic techniques satisfying the estimated influential factors for a successful enterprise.
A case study is applied in Smart Village, Cairo, Egypt, to show the applicability of the proposed model. The
smart village’ consistency rate is measured after applying neutrosophic methodologies to reach to nearest
optimum results. Additional case studies on the smart city in the U.K. and China have been presented to
justify that our proposal can be used and replicated in different environments.

INDEX TERMS Multi-criteria decision making (MCDM), analytical hierarchal Process (AHP),
neutrosophic sets, Internet of Things (IoT).

I. INTRODUCTION the successful deliveries of IoT enterprise. The research esti-

IoT, widely regarded as a novel engine for information and mates an IoT framework for small and medium enterprise.
communications technology industry, was estimated to lead Based on literature review and expert interviews, five major
market within next ten years [1]. The ramification of IoT influential factors have been detected. The five major influ-
on consumer and technical sectors make the extraordinary to ential factors are security, value, connectivity, intelligent, and
reform industry revolution. IoT merges the power of internet telepresence as follows [4]:
with the competence of industries to conduct real world of
factories, machine, goods, and infrastructure [2]. IoT empow- 1) Security: The right information can be integrated with
ers the control of things (networks, desktop, laptops, etc.) specific legislation to restrict handling of IoT mecha-
to ensure the delivery of perfect and smart enterprise, and nisms and rules.
to develop IoT products or services all over the world [3]. 2) Value: The benefits that can impact on the attitude and
Mainly the current challenges face enterprises are security the manner of behavior according to enterprises.
issues such as lack of confidence and understanding of IoT. 3) Connectivity: Backend systems behind IoT objects
Although IoT has positive effects on enterprises, it also has are vital to maintain keep smooth communications and
many negative impacts to be reduced or removed to guarantee successful deliveries offered by applications. The mean
of connectivity in the proposed study is to keep all
The associate editor coordinating the review of this manuscript and objects and people connected with the capabilities and
approving it for publication was Guangjie Han. technologies of IoT.
2169-3536
2019 IEEE. Translations and content mining are permitted for academic research only.
VOLUME 7, 2019 Personal use is also permitted, but republication/redistribution requires IEEE permission. 59559
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
 N. A. Nabeeh et al.: Neutrosophic Multi-Criteria Decision Making Approach for IoT-Based Enterprises

4) Intelligent: IoT devices have a feature of intelligence combine various decision makers’ perspectives to achieve the
to differentiate the usual Internet from IoT devices. ideal perspectives by handling the confliction and biasness
Also IoT machines can intelligently receive input infor- between decision makers. To ensure the effectiveness of the
mation and produce instructions in order to complete model proposed, an efficient case study is applied to smart
task. city Cairo, Egypt. In addition, a validation of case studies
5) Telepresence: The connections between different in UK and China is presented to ensure the replication of the
objects on internet via wireless technology can allow proposed model.
meetings without physical attendance. The reliable Section 2 mentions literature review of the current knowl-
IoT products give consumer positive impression for the edge include methodological contributions have been pre-
service. sented from other researchers. Section 3 presents some basics
MCDM can be referred as a formal and structured definition for neutrosophic environment. Section 4 illustrates
decision making methodology for dealing with complex methodology of the proposed model and the way to help
problems and conflicting criteria [5]. Nowadays AHP is the decision makers in the estimation of the influential fac-
most widespread method deals with MCDM problems [6]. tors affecting the success of enterprise. Section 5 confirms
AHP allows the use of both quantitative and qualitative cri- the validity of proposed model by presenting a case study.
teria in evaluation. AHP basic steps are concluded in three Section 6 applies validation for the proposed model in UK
consecutive steps which are decomposition, calculation of and China. Section 7 concludes the research and points to the
decision criteria weights, and calculate priorities of the candi- future of the work of research.
date’s alternatives [7]. Business environments can be threat-
ened by uncertainties. The uncertain circumstances would II. LITERATURE REVIEW
force researchers to monitor and to manage the estimated The method of MCDM, become a strategic issue for multiple
misjudgment induced from uncertainty [8]. IoT applications, decision makers in organizations, is developed for the selec-
such as enterprise, marketing, healthcare, decision theory, and tion process with ordinal preferences of criteria and alterna-
finance can be accelerated by the surrounding of influential tives. In [16], a case study is developed to MCDM considering
factors [9]. Classical AHP can detect priorities for candi- the weights of criteria and decision makers. The globalization
date’s criteria in addition can compare, and rank alternatives. becomes an essential strategic decision power in the selection
The classical AHP cannot deal with impression and vague problems, the use of AHP perceived as an effective tool
information. In addition, the saaty comparison matrix has to be tackled. In [17], a case study developed a model to
no systematic methodology to detect whether the matrix is solve the selection problems using AHP methods. In [17],
inconsistent state or not. The AHP using Fuzzy approach uses the techniques of AHP to assist the MCDM problems
has the same advantages of classical AHP in addition to by comparing the weights of the summation of number of
dealing with vague or imprecise through one grade. Fuzzy rank vote. The research of [18] uses AHP to solve MCDM
AHP deals with membership function to detect preference problems in order to achieve to the best solution of candidates
relations [7]. Due to environment constraints, decision mak- cloud services based on quality of service attributes. The
ers cannot consistently detect the membership function. researchers propose to use AHP methods in order to gen-
To overcome current challenges of MCDM methods, erate weights of the problem [19]. Researchers propose an
the MCDM is combined with fuzzy approaches to esti- AHP method to rate and select the appropriate suppliers with
mate possible solutions to grant enterprise successful as respect to evaluating criteria [17], [20]–[27]. The use of AHP
mentioned [10], [11]: in MCDM problems can be used to solve quantitative and
1) The existence of various and conflicting criteria, and qualitative problems, for obtaining the related alternatives,
alternatives. criteria, and sub criteria [28].
2) Decision maker’s different perspectives and interests. To overcome the classical challenges of AHP methods of
3) Process of estimation to best criteria usually has vague relying on impression and vague information, the challenge
and impression information. of the existence of multiple decision makers, alternatives, and
4) Decision makers must have a great magnitude of cog- criteria, a fuzzy multi-criteria analysis framework is proposed
nitive in order to achieve optimal estimation under to evaluate the performance of IoT in specific field of enter-
difficult circumstances [12]–[14]. prises. The intuitionistic fuzzy is used to handle the vague
The Neutrosophic sets model real world problems with and impression of the evaluation process [29]. The evolution
respect to the conditions of all decision making situa- of fuzzy multi-criteria group decision making model affords
tions [15]. This research illustrates AHP methods combined enterprises capabilities to appraise the performance of the
with neutrosophic techniques to be effectively present the cri- IoT supply chain. Fuzzy and AHP methods are applied on
teria related to influential factors. Our proposed model helps a rule based decision support mechanism for evaluating the
decision makers to professionally estimate the influential fac- IoT influential factors [4] The expansion of classical AHP
tors to ensure success of related IoT services. The proposed with fuzzy methods is convenient with MCDM environment.
model can efficiently deal with uncertain and inconsistent The fuzzy preference programming (FPP) reveals that the
information by the use neutrosophic set. In addition, we can used weights cannot present the actual relations between

59560 VOLUME 7, 2019

N. A. Nabeeh et al.: Neutrosophic Multi-Criteria Decision Making Approach for IoT-Based Enterprises

alternatives and criteria, and the existence of confliction no restriction on the sum of TNe (x) INe (x), and FNe (x),
between criteria, which leads to a logarithmic fuzzy prefer- so 0− ≤ sup TNe (x) + sup INe (x) + sup FNe (x) ≤ 3+ .
ence programming (LFPP) using the priority of the deviation Definition 2: A single valued neutrosophic set Ne over X
of Fuzzy AHP [30]. Authors in [31]–[33] mentioned a hier- taking the following form X = {hX , TNe (x), INe (x), FNe (x)i :
archy model combined with fuzzy sets to solve the problems X ∈ X }, where TNe (x) : X → [0,1], INe (x) : X → [0,1] and
of selection. The linguistics terms are used to assess the FNe (x) : X → [0,1] with 0 ≤ TNe (x) + INe (x) + FNe (x) ≤ 3
weights and to rate the evaluating factors. In [32] numerous for all X ∈ X . The single valued neutrosophic (SVN) number
researchers mention a systematic review of literature of the is symbolized by Ne = (d, e, f ), where d, e, f ∈ [0, 1] and
MCDM approaches for selection. In [33], [34] MCDM tech- d + e + f ≤ 3.
niques illustrate how to overcome multiple, and conflicting Definition 3: The single valued triangular neutrosophic
objectives using fuzzy principles. In [35], illustrates fuzzy number, a = ((a1 , a2 , a3 ) : αa , θa , βa ) is a neutrosophic set
techniques for decision making to ensure achieving ideal on the real line set R, whose truth, indeterminacy and falsity
decision with respect to different criteria and condition of membership are as follows:
market. The growth of shopping centers and business cen-  
x − a1

ters makes researchers find a way to view recommendation

 α a (a1 ≤ x ≤ a2 )
 a2 − a1
factors, which appear to be easier and more accessible than Ta (x) = (1)
αa (x = a2 )
those by traditional ways. Intelligent interactive marketing



IoT systems could perform effective ways between service 0 otherwise,
(a − x + θ (x − a1 )

providers and consumers [36]. In [37], a self-organized IoT 2 a

 (a1 ≤ x ≤ a2 )
aware system illustrated for online shopping by aggregating  (a2 − a1 )
all possible preferences. In [4], illustrates a rule-based deci- Ia (x) = θ (x = a2 ) (2)

 a
sion support system for IoT enterprise using fuzzy to detect 
1 otherwise,
the influential factor affected the success of IoT-enterprise 
(a2 − x + βa (x − a1 )
The AHP methods combined with fuzzy techniques can

 (a1 ≤ x ≤ a2 )
(a2 − a1 )


work with vague information but it is not the best way

Fa (x) = βa (x = a2 ) (3)
forward [40]. In [41], MCDM procedures are proposed via 
 (x − a 2 ) + β a (a3 − x))
neutrosophic sets to deal with inconsistent and uncertain (a2 ≤ x ≤ a3 )



cases. An approach in [42] is used to predict cloud services (a3 − a2 )
qualification. The use of triangular neutrosophic numbers aid where, αa , θa , βa ∈ [0, 1] and a1 , a2 , a3 ∈ R, a1 ≤ a2 ≤ a3 .
to work on inconsistent and ambiguous information. An effi- Definition 4: Let a = h(a1 , a2 , a3 ) ; αa , θa , βa i and b =
cient model is used to estimate solutions for estimation obsta- h(b1 , b2 , b3 ) ; αb , θb , βb i be two single valued triangular neu-
cles. The indeterminate and inconsistent data is powerfully trosophic numbers and γ 6= 0 be any real number. Then,
handled using the neutrosophic sets by considering the level 1. Addition of two triangular neutrosophic numbers
of truth, indeterminate, and false degrees. In [43] a general a + b = h(a1 + b1 , a2 + b2 , a3 + b3 ) ; αa ∧ αb , θa
framework uses a single valued neutrosophic and rough set
∨θb , βa ∨ βb i
theories to handle the uncertainty and inconsistency. The pro-
posed method improves the decisions and service in the use 2. Subtraction of two triangular neutrosophic numbers
of IoT in smart city. The neutrosophic theory is an effective
a − b = h(a1 − b3 , a2 − b2 , a3 − b1 ) ; αa ∧ αb , θa
method for dealing with inconsistent data, the three ways
decision according to neutrosophic set is proposed to achieve ∨θb , βa ∨ βb i
a reasonable effective decisions [44]. A neutrosophic three 3. Inverse of a triangular neutrosophic number
membership functions proposed to support the calculation
of weights corresponding to alternatives and criterions for a−1  
1 1 1
choosing the most appropriate alternative. The effective alter- = , , ; αa , θa , βa , where (a 6= 0)
native resulted will improve quality of service, in addition a3 a2 a1
will make a well-defined reduction in cost, and time. 4. Multiplication of triangular neutrosophic number by
constant value
(
III. BASIC DEFINITIONS OF NEUTRSOPHIC SETS h(γ a1 , γ a2 , γ a3 ) ; αa , θa , βa i if (γ > 0)
In this section, important definitions of neutrosophic set are γa =
h(γ a3, γ a2 , γ a1 ) ; αa , θa , βa i if (γ < 0)
clearly [42], [44]:
Definition 1: The neutrosophic set N characterized by three 5. Division of triangular neutrosophic number by con-
membership functions which are truth-membership func- stant value
  
tion TNe (x), indeterminacy-membership function INe (x) and a1 a2 a3
, γ , γ ; αa , θa , βa if (γ > 0)


=  aγ
falsity-membership function FNe (x), where x ∈ X and X a 
be a space of points. Also TNe (x) : X →]− 0, 1+ [, INe (x) :
 
γ  3 a2 a1
, , ; αa , θa , βa if (γ < 0)
X →]− 0, 1+ [and FNe (x) : X →]− 0, 1+ [. There is γ γ γ



VOLUME 7, 2019 59561

 N. A. Nabeeh et al.: Neutrosophic Multi-Criteria Decision Making Approach for IoT-Based Enterprises

The importance of IoT has been suggested by numerous

of researchers for the connection of internet based between
different devices [1], [2]. However, the process of decision
making for IoT-enterprise faces the conditions of uncer-
tainty, and inconsistency of data. The interconnected devices
of IoT makes the opportunity of accompanied many influ-
ential issues that affecting the performance of enterprise.
Fig. 2 presents the enterprise’s model and the corresponding
hierarchal levels. We present the main enterprise hierarchal
levels which are [45]:
• Strategic Level: the top managers achieve the strate-
gic enterprise goals by operating some activities and
judgments, indeed the strategic polices will influence
FIGURE 1. Influential Factors for IoT enterprise adoption. on the success and performance of enterprises. The
environment in strategic level is under the conditions of
uncertainty.
6. Division of two triangular neutrosophic numbers • Tactical Level:the middle managers develop planning to
   achieve objectives and targets of strategic levels. Since
a1 a2 a3

 ,
b3 b , ; α a ∧ α ,
b aθ ∨ θ ,
b aβ ∨ β b characteristics of decisions in strategic level are taken
2 b1



 generally for whole enterprise, the decisions in tactical
if (a3 > 0, b3 > 0)




   levels would be clearer than in strategic levels. The way
 a 3 a 2 a1 makes decisions be more rapid and customized.
, , ; αa ∧ αb , θa ∨ θb , βa ∨ βb

a 
= b3 b2 b1 • Operational Level:the decisions in operational level
b  

  if (a3 <0, b3 > 0)  deal with daily operations to complete the vision and
a a 2 a1

 3 strategy of strategic and tactical levels. The implemen-
, , ; αa ∧ αb , θa ∨ θb , βa ∨ βb



 b3 b2 b1

 tation of operational levels can be performed by enter-
if (a3 < 0, b3 < 0)

 prise’s junior managers.
7. Multiplication of two triangular neutrosophic numbers The hierarchal process transforms data from loose irrel-
evant data to useful information, reaching knowledge and
h(a b , a b , a b ) ; αa ∧ αb , θa ∨ θb , βa

final levels for decision. Enterprise model illustrates the
 1 1 2 2 3 3




 ∨βb i if (a3 > 0, b3 > 0) competitive strategy, enterprise strategy, and enterprise
structures [45], [46]. The enterprise structure includes differ-

h(a b , a b , a b ) ; α ∧ α , θ ∨ θ , β

1 3 2 2 3 1 a b a b a
ab = ent applications of inventory, facilities, sales, sourcing, and

 ∨β b i if (a3 < 0, b3 > 0) others. Fig. 3. Represents details about IoT structure [1], [47]:

h(a3 b3 , a2 b2 , a1 b1 ) ; αa ∧ αb , θa ∨ θb , βa


• IoT Formal Definition: interconnected objects over


∨βb i if (a3 < 0, b3 < 0)


network without human intervene. Anybody can access
from anyplace the required content to achieve personnel,
IV. RESEARCH METHODOLOGY business, or medical tasks.
Saaty [6] illustrates the AHP as a widespread multi-criteria • Enterprise IoT Applications: enterprises gain great
decision making technique for efficient decision making. benefits to merge the current applications with the tech-
AHP can be an imperative method for managers to solve nologies of IoT. The enterprise IoT applications are
complex and confusion problems. The AHP decomposes supply chain, connected cars, retail, farming and others.
problems to sub-problems for the purpose of simplicity. • IoT Reference layers: The IoT architectures can be built
AHP is imperative method for mangers to solve complex and by the use of reference layers and definitions.
confusion problems. The problem criteria can be calculated Fig. 4. combines enterprise with IoT which evolves the
by using the pair-wise comparison judgment. Neutrosophic need for novel techniques of decision making. The challenges
set is integrated with AHP technique; the relative significant face the traditional decision makings problems of uncertainty,
factors are scaled by neutrosophic ratio. The relative effec- inconsistency, vague, and impression. The traditional steps
tives of criteria indicated using neutrosophic numbers. The of decision making can identifies problems in enterprise,
proposed study illustrates the influential factors affecting the identifies the surrounding criterions, performs priorities for
success of organization as mentioned in Fig.1. The applica- set of available alternatives, and evaluates the efficiency
tions of IoT used in enterprises are collecting data from dif- of decisions. Absolutely, the traditional steps for decision
ferent private or public domains [4]. The interconnected IoT making cannot handle the current challenges. So there is a
devices can provide better opportunities, in both technical and necessary to combine neutrosophic theory to enhance the
business aspects, which have direct impacts on enterprises. performance of decision making process. The main steps of

59562 VOLUME 7, 2019

N. A. Nabeeh et al.: Neutrosophic Multi-Criteria Decision Making Approach for IoT-Based Enterprises

FIGURE 2. Enterprise Models and hierarchical levels for obtaining final decision.

FIGURE 3. IoT definitions, applications, and IoT reference models.

the proposed methodology are presented in Fig. 5, and the alternatives via linguistic terms. The decision maker
detailed descriptions are as follows: presents that criteria 1 is strongly important than criteria 2.
Step 1: Determine the objective of your study; decompose The triangle neutrosophic scaled as h(2, 3, 4) 0.40, 0.65, 0.60i.
problem hierarchy to represent the goal, criteria, and the Conversely, if the decision maker presents that criteria 2
possibility of alternatives. is slightly significant than criteria 1, then the
Step 2: Decision makers use neutrosophic scale pre- triangle neutrosophic scale would be as 1/h(2, 3, 4) 0.40,
sented in table 1 to make comparison between criteria and 0.65, 0.60i.

VOLUME 7, 2019 59563

 N. A. Nabeeh et al.: Neutrosophic Multi-Criteria Decision Making Approach for IoT-Based Enterprises

FIGURE 4. Enterprise based IoT for successful decision making.

FIGURE 5. The neutrosophic AHP steps for successful IoT Enterprise.

The following form 4 presents the pairwise matrices of where rijk represents the k th decision maker based on
comparing different criteria with each other the relation of preference of ith over jth criteria. The
triangular neutrosophic scale is in the form of rijk =
 
k
r11 . . . r1nk

Ak =  ... .. .. 
D  E

. .  (4) lijk , mkij , ukij ; Tijk , Iijk , Fijk , Such that lijk , mkij , ukij are the
k
rn1 · · · rmn k lower, median and upper bound of neutrosophic number,

59564 VOLUME 7, 2019

N. A. Nabeeh et al.: Neutrosophic Multi-Criteria Decision Making Approach for IoT-Based Enterprises

TABLE 1. THE triangular neutrosophic scale of AHP. TABLE 2. Upper bound of paire-wise comparison matrix.

1. Calculate the average row:

n
P
(xij )
j=1
wi = ; i = 1, 2, 3, . . . .m; j = 1, 2, 3, . . . , n
n
(8)

2. The normalization of crisp value is calculated using the

following equation
wi
wm
i = P m ; i = 1, 2, 3, . . . , m. (9)
wi
i=1
Step 5: Check the consistency of decision makers of
judgments
Transitive is used to determine the consistency of judg-
Tijk , Iijk , Fijk ’’ are the truth-membership, indeterminacy and ments matrix .Such that if the pair-wise comparison has a
falsity membership functions respectively of triangular neu- transitive relation i.e. aik = aij ajk for all i, j, andk, then
trosophic number. For instance, r21 3 is the preference relation a pair-wise comparison matrix considered to be consis-
of second criteria and first criteria, corresponding to the tent. Therefore
relation i.e. (lik , mik , uik ) =
a transitive
lij , mij , uij . ljk , mjk , ujk is proposed to detect the consis-


third decision makers and has the following neutrosophic
scale:r21 3 = h(6, 7, 8) ; 0.90, 0.10, 0.10i. tency. The consistency rate (CR) is very important for calcula-
Step 3: considering not only one decision maker to esti- tions, since CR is the computed ratio between the consistency
mate the preferences between relations, the aggregated rij as index (CI) and a random consistency index (RI). The rate of
follow. (CR) cannot be more than 0.1 with respect to comparison
k D  E matrix, such that the proposed matrix is less than or equal
lijk , mkij , ukij ; Tijk , Iijk , Fijk to 4×4 . If upper bound of the CR for the proposed matrix
P
k=1 illustrated as shown in table 2 [42], the matrix is state of
rij = (5)
k inconsistence.
The average values for the estimated preferences are cal- The following steps show the calculation of CI and CR:
culated via the aggregated pair-wise comparison matrix as 1. The comparison matrix’s columns are multiplied by
follows: its corresponding priority. The summation of all rows
resulting of values in form of vector called ‘‘weighted
r11 . . . r1n
 
sum’’.
A =  ... ..
.
.. 
.  (6) 2. The values for weighted sum vector are divided by each

criteria’s equivalent priority
rn1 ··· rmn
3. Calculate the mean for the preceding step values stands
Convert the neutrosophic scales to crisp values by apply score for λmax .
functions of rij as mentioned in [48]: 4. The consistency index (CI) is computed as mentioned:
  λmax − n
 Tij + Iij + Fij  CI = , (10)
s(rij ) = lij × mj × uij )
 (7) n−1
9 
where n is the number of the compared criteria.
where l, m, u denotes lower, median, upper of the scale 5. Compute the consistency ratio, which is defined as:
neutrosophic numbers, T, I, F are the truth-membership, inde-
terminacy, and falsity membership functions respectively of CI
, CR = (11)
triangular neutrosophic number. CR
Step 4: Based on the preceding matrix, weights and prior- where RI is the random produced matrix consistency index
ities are calculated as presented and illustrated in table 3.

VOLUME 7, 2019 59565

 N. A. Nabeeh et al.: Neutrosophic Multi-Criteria Decision Making Approach for IoT-Based Enterprises

TABLE 3. Saaty table for random consistency index (RI) per different TABLE 4. Main five variable’s operational definitions.
number of criteria.

Step 6: Decision makers who could make repetitive exer-

cise in case of inconsistency of matrix of classical AHP.
In neutrosophic AHP, only decision makers involved in
repairing the pair-wise comparison matrix could improve
consistency degree by following to the next steps. To ensure
the consistency, the inconsistent elements should be selected
on the pair-wise comparison matrix using the induced matrix
illustrated in [42]. The theorem and corollaries can be used as
mentioned in [49]. Major steps are used to identify inconsis-
tency of pair-wise comparison matrix to improve the degree
of consistency rate mentioned:
1. Formulate the neutrosophic induced matrix
I =A×A−n×A
TABLE 5. Information about decision makers.
2. Detect the largest preference relation r̃ij such that has
the largest lower, median and upper-bound of triangular
number.
3. Detect the ith row and jth column which encompass
inconsistent triangular neutrosophic number. Compute
dot product of row vector Roi = (ri1 , ri2 , . . . , rin ) and
column vector CoTj = (r1j , r2j , . . . , rnj ), where CoTi is
the transpose vector of Coj .
4. The dot product
makers cannot detect the impact of related consequences.
P = Roi .CoTj = (ri1 r1j , ri2 r2j , . . . , rin rnj ) (12) The influential factors of IoT enterprise are security, value,
5. Compute elements far from rij in vector P according to connectivity, intelligent, and telepresence which presented
the mentioned formula: in table 4. The enterprise needs to make evaluation of influ-
ential factors in order to insure good IoT connectivity system
b = P − rij (13) and to attain a successful IoT-related enterprise. The IoT
such that P is the prejudice vector. enterprise alternatives for using of big data tools for are
6. Use prejudice to detect inconsistency by modifying (1) Spark, (2) KNIME, and (3) Hadoop. The five criteria
element A of original pair-wise comparison matrix’s in for enterprise decision makers are (1) security, (2) value,
element. (3) connectivity, (4) telepresence, and (5) intelligent.
7. The inconsistent elements are defined to be the largest Step 1: Draw the hierarchy of IoT influential factors of
lower, median and upper bounds in addition to be far enterprises process as in Fig 6, and mention information about
from scratch in the prejudice vector. decision makers and interviewers as mentioned in table 5.
8. In order to reach to the consistency of judgments the Step 2: A session has been performed with strategic level
inconsistent elements must be modified of enterprise directors and decision makers in order to make
Step 7: Calculate the normalized weights of alternatives as comparisons and average preferences between criterions and
in criteria weight calculation process. An alternative score can alternatives using neutrosophic scales in table 1.
be achieved by multiplying each alternative to its correspond- Step 3: an aggregated pairwise comparison matrix rep-
ing weight with respect to corresponding criteria resents the average preferences and judgments of decision
Step 8: Rank alternatives according to highest score value. makers and, modeled in the form of neutrosophic scales as
mentioned in table 6. For sake of simplicity, the aggregated
V. THE NEUTROSOPHIC AHP DECISION SUPPORT FOR pair-wise comparison matrix has been converted into crisp
IOT INFLUENTIAL FACTORS OF ENTERPRISE values using Eq. (7) and results represented in table 8.
The proposed case study has been applied on smart village big Step 4: Compute the criteria’s weight
data in Egypt. A smart village enterprise exposes some com- 1. Calculate the average of row using the presented using
mon characteristics to delivers insight to customers. Although Eq. (8) w1 = 1.6202 w2 = 1.4888 w3 = 1.0986 w4 =
smart applications pioneered by enterprises, but decision 0.9096 w5 = 0.623.

59566 VOLUME 7, 2019

N. A. Nabeeh et al.: Neutrosophic Multi-Criteria Decision Making Approach for IoT-Based Enterprises

TABLE 6. Neutrosphic pair-wise comparison matrix of criteria.

TABLE 7. Crisp values of judgments of neutrosophic pair-wise matrix.

TABLE 8. The comparison matrix of criteria after modification.

2. The normalization illustrated to normalize the crisp 2. Divide the weighted sum vector’s value by the criteria’s
value, the criteria’s corresponding normalized weights corresponding priority as follows: w1 = 5.482 w2 =
mentioned using Eq. (9): w1 = 0.282 w2 = 0.259 w3 = 5.038 w3 = 4.990 w4 = 4.810 w5 = 5.326.
0.19 w4 = 0.15 w5 = 0.10. 3. Calculate the average of the preceding step results
P
It’s obvious that wi = 1. which is stand for λmax , then λmax = 5.1295.
The arrangement of criteria with respect to priorities is C1 , Since λmax still neutrosophic number, then apply de-
C2 , C3 , C4 and C5 respectively. neutrosophic as mentioned
Step 5: Check consistency of judgments. 4. Calculate the consistency index (CI) as mentioned:
The pair-wise comparison matrix is consistent if and λmax − n 5.1295 − 5
only if there exist a transitive relation such aik = CI = = = 0.03,
n−1 4
aij ajk foralli, j, and k. The consistent degree is calculated as
illustrated in next steps: where n represent the number of proposed criteria.
5. Calculate the consistency ratio as illustrated:
1. Compute the ‘‘weighted sum’’ for each row w1 =
1.547 w2 = 1.306 w3 = 0.955 w4 = 0.762 CI 0.03
CR = = = 0.02
w5 = 0.578. RI 1.12

VOLUME 7, 2019 59567

 N. A. Nabeeh et al.: Neutrosophic Multi-Criteria Decision Making Approach for IoT-Based Enterprises

FIGURE 7. The evaluation criteria performed weights.

FIGURE 6. AHP decision support for IoT influential factors of enterprises.

Consistency rate is optimized near to 0.1, with respect to
saaty as mentioned in table 2. It is apparent that when CR
became close to 0.1, we were able to reduce the consistency
Step 6: Since the proposed pair-wise comparison is 5 x
rate CR from 0.02 to 0.01. The resulted CR is considered to be
5, then CR must be less than 1.12 as illustrated in table 2,
efficient with the comparison of the value 1.12 as mentioned
the resulting CR is an appropriate ratio to the comparison
in table 2.
matrix. However, we can enhance the resulted CR ratio to be
The proposed criteria examined for its applicability and
near to 0.1 in order to achieve the high degree of consistency.
benefits using four criteria proposed in [40]
1. Create the induced matrix I = A.A − n.A.
• Correlation: interdependency between criteria showed
−0.1 −2.9 1.72 0.57 4.485 using the correlation coefficient of Spearman as
1.795 −0.02 −2.225 −1.34 2.05 mentioned:
−0.2 1.28 −0.01 −0.61 −0.545
6 ∗ D2i
P
0.25 0.73 0.42 −0.1 −2.32 ρ =1− (14)
−0.706 0.09 0.41 1.195 −0.01 n ∗ (n2 − 1)
where ρ is the correlation coefficient, Di = xi − yj
2. The largest preference relation r15 .
represent the difference between the value of ranked
3. The dot product P =Ro1 · CoT5 = Ro1 = (1, 1.8488,
criteria values, such that n is the number of criteria.
1.38, 2.03, 1.843) CoT5 = (1.843, 2.03, 1.843,
There is a strong correlation between criterions as shown
1.848, 1) P = (1.843, 3.7, 2.5, 3.7, 1.843)
in the following computations:
4. Compute elements that far from rij in vector P accord-
ing to the mentioned formula: ρSecurity−Value = 0.99,
P − r15 = (0, 1.875, 0.657, 1.857, 0) ρSecurity−Intelligent = 0.92,
ρSecurity−Telepresence = 0.93,
5. The consistent elements in b is all elements that contain ρSecurity−Connectivity = 0.96
rather negative or zero values other elements are needed
to be enhanced. ρValue−Intelligent = 0.91,
6. The comparison matrix’s consistency is enhanced by ρValue−Telepresence = 0.99,
modifying r15 as mentioned in table 8. ρValue−Connectivity = 0.98,
The normalized weight values of the preceding matrix in ρConnectivity−Intelligent = 0.90,
table 8 will be as mentioned: w1 = 0.260 w2 = 0.267
w3 = 0.19 w4 = 0.16 w5 = 0.11 ρConnectivity−Telepresence = 0.94,
The priorities of criteria are presented in Fig.7 as follows: ρIntelligent−Telepresence = 0.97.
C2 , C1 , C3 , C4 and C5 respectively so that, security and
Step 7: compute alternative’s weights with respect
value are the most important criteria according to company’s
to criteria.
directors.
Repeat the de-neutrosophic process to neutrosophic scales
By computing λmax as we mentioned previously with
into crisp values by the use of Eq. (7), use the methods
details, we found that λmax = 5.07 Compute the consistency
of calculation of weights of criteria, and then compute the
index (CI) as follows:
alternative’s normalized weight as mentioned:
λmax − n 5.07 − 5 • The alternatives of comparison matrix with respect to
CI = = = 0.017
n−1 4 security criteria are mentioned in table 9. Such that,
CI 0.017 A1 , A2 and A3 are corresponding to Spark, KNIME and
CR = = = 0.01
RI 1.12 Hadoop respectively.
59568 VOLUME 7, 2019
N. A. Nabeeh et al.: Neutrosophic Multi-Criteria Decision Making Approach for IoT-Based Enterprises

TABLE 9. The alternatives of pair-wise comparison matrix according to security.

TABLE 10. The alternatives of pair-wise comparison matrix according to value.

TABLE 11. The alternatives of pair-wise comparison matrix according to connectivity.

1. Calculate the average of row using the presented • The alternatives of comparison matrix with respect to
Eq. (8): connectivity criteria have been mentioned in table 11.
The connectivity criteria and its corresponding alter-
w1 = 1.624 w2 = 0.974 w3 = 0.736
natives of normalized weights are mentioned as
2. The normalization illustrated to normalize the mentioned:
crisp value, the criteria’s corresponding normal- 1. Calculate the average of row using the presented
ized weights mentioned using Eq. (9): Eq. (8):
w1 = 0.48 w2 = 0.29 w3 = 0.22 w1 = 1.624 w2 = 1.11 w3 = 0.694
• The alternatives of comparison matrix with respect to 2. The normalization illustrated to normalize the
value criteria have been mentioned in table 10. crisp value, the criteria’s corresponding normal-
The value criteria and its corresponding alternatives of ized weights mentioned in Eq. (9):
normalized weights are mentioned as mentioned:
1. Calculate the average of row by the use of the w1 = 0.47 w2 = 0.32 w3 = 0.20
presented Eq. (8): • The pair-wise comparison matrix of alternatives with
respect to intelligent criteria shown in table 12. The
w1 = 1.47 w2 = 1.189 w3 = 0.677
Intelligent criteria and its corresponding alternatives of
2. The normalization illustrated to normalize the normalized weights are mentioned as follows:
crisp value, the criteria’s corresponding normal- 1. Calculate the average of row by the use of the
ized weights mentioned in Eq. (9): presented Eq. (8):
w1 = 0.44 w2 = 0.35 w3 = 0.209 w1 = 1.47 w2 = 1.111 w3 = 0.677

VOLUME 7, 2019 59569

 N. A. Nabeeh et al.: Neutrosophic Multi-Criteria Decision Making Approach for IoT-Based Enterprises

TABLE 12. The alternatives of pair-wise comparison matrix according to intelligent.

TABLE 13. The alternatives of pair-wise comparison matrix according to telepresence.

2. The normalization illustrated to normalize the

crisp value, the criteria’s corresponding normal-
ized weights mentioned in Eq. (9):
w1 = 0.45 w2 = 0.34 w3 = 0.20
• The pair-wise comparison matrix of alternatives
with respect to telepresence criterion is presented
in table 13. The telepresence criteria and its correspond-
ing alternatives of normalized weights are mentioned as
mentioned:
1. Calculate the average of row by the use of the
presented Eq. (8): FIGURE 8. Comparison of three alternatives according to different
criteria.
w1 = 1.47 w2 = 1.187 w3 = 0.75
2. The normalization illustrated to normalize the Step 8: Rank the recommended alternatives according to
crisp value, the criteria’s corresponding normal- highest score value as mentioned in Fig 12.
ized weights mentioned in Eq. (9):
w1 = 0.43 w2 = 0.34 w3 = 0.22 VI. VALIDATION IN THE UK AND CHINA
It is imperative for the working proposal to be validated
The weight of three alternatives of the smart village accord-
in different context, including different countries, institu-
ing to each criterion is mentioned in Fig.8. For sake of
tions and sectors. Without any exception, our proposal has
description, Fig 9, 10, and 11 present a detail analysis for each
been validated in the UK and China to ensure that it can
alternatives with respect to the related criteria.
be replicated, reusable and adaptable. We follow the steps
Multiply each criterion by its corresponding weights to
described between Section 3 and Section 6. We also interview
obtain the score value
five representatives in the UK and five representatives in
The alternatives relative score value is as mentioned:
  China to make comparative studies and understand any differ-
0.48 0.44 0.47 0.45 0.43
 0.29 0.35 0.320.340.34  ences due to different locations, cultures and emphasis. Each
representative presents the core values for each business.
0.22 0.20 0.20 0.20 0.22
  We focus on results similar to between Fig.9 and 12. We can
0.260   then successfully analyze rational behind.
 0.267  0.45 Fig.13 shows results for comparison of Spark alternative
 
×  0.19  = 0.32 according to different criteria in the UK. All these five rep-
  
 0.16  0.20 resentatives have similar values and rating scores under 0.5,
0.11 since they believe that maintaining a good balance in all the
59570 VOLUME 7, 2019
N. A. Nabeeh et al.: Neutrosophic Multi-Criteria Decision Making Approach for IoT-Based Enterprises

FIGURE 9. Comparison of Spark alternative according to different criteria. FIGURE 12. The priorities of smart village alternatives with respect to
related criteria in Egypt.

FIGURE 10. Comparison of KNIME alternative according to different

criteria in Egypt.
FIGURE 13. Comparison of Spark alternative according to different
criteria in the UK.

FIGURE 11. Comparison of Hadoop alternative according to different

criteria in Egypt.
FIGURE 14. Comparison of Hadoop alternative according to different
criteria in the UK.
factors are necessary. Even if the levels of competitions can
be high and the extents of uncertainty can be volatile, the best
approach for them is to maintain all key factors smartly stable scale of deployment and services due to the demands on IoT,
and steady, rather than being excellent in one or two factors. Edge Computing and AI. Connectivity has been expanded
Even so, Intelligent has the highest scores and the value has on connecting different smart cities, smart services, smart
the lowest scores even the differences are not far. This is devices and smart robots, particularly in London. Therefore,
because services should be adaptable to meet market demands the scores for Intelligent and Connectivity are higher than the
and customers’ requests. other three, which have the same score of 0.4 each.
Fig. 14 shows results for comparison of Hadoop alternative Fig.15. shows results for comparison of Spark alterna-
according to different criteria in the UK. All the scores are tive according to different criteria. Connectivity is the most
below 0.5, but are more well-balanced since these five rep- important criteria as reflected by five Chinese representative
resentative firms consider they are all important. Intelligent firms since all services and users must be online and con-
and connectivity are considered the most important criteria nected. In China, there are millions of users. Disconnecting
as follows. First, a lot of services have been completed by from any services, business transactions and online visits may
Hadoop. More requests have been made about increasing the result in millions of financial loss. Due to the restrictions in

VOLUME 7, 2019 59571

 N. A. Nabeeh et al.: Neutrosophic Multi-Criteria Decision Making Approach for IoT-Based Enterprises

FIGURE 15. Comparison of Spark alternative according to different

criteria in China. FIGURE 17. The priorities of smart village alternatives with respect to
related criteria in London.

FIGURE 16. Comparison of Hadoop alternative according to different

criteria in China. FIGURE 18. The priorities of smart village alternatives with respect to
related criteria in Shanghai.

some security setting, then connectivity can only go for 0.8 at

most. The other scores are as low as between 0.2 and 0.3. ‘‘others’’. Figure 17 and 18 show the priorities of smart
Fig.16 shows results for comparison of Hadoop alterna- village alternatives with respect to related criteria in London
tive according to different criteria. It has a similar shape and Shanghai respectively. Both are big cities and thus their
like Fig.15, except it has higher scores for value, intelligent, orientation is presented as the smart city. In London, others
and telepresence. In other words, it means Hadoop services consist of 45%; Hadoop has 34% and Spark has 21% of
are more mature and more established than Spark services. percentage of usage and deployment. There is a trend that
Hadoop was in used by IoT services earlier than Spark. others may still go up, since there are more varieties of
However, security still remains challenges for IoT services different solutions on offer.
in China. Fig.18 shows interesting results. Alibaba is one of the
Comparing between services in the UK and China, we can biggest IT service providers in China. Hence, the differ-
identify that UK service providers and users are more con- ence is there are Spark and Hadoop services offered by
cerned that services should be well-balanced in all important Alibaba or non-Alibaba. Continentally, it has 20% each for
criteria. Differences between them are smaller. Whereas in Spark and Hadoop services by Alibaba (Ali) and 20% each
China, the most important factor is the connectivity to ensure for Spark and Hadoop services by non-Alibaba services. The
all payment and business transactions can be made efficiently remaining 20% is for all other services not using Spark and
and quickly. Millions of financial transactions can be made Hadoop. Shanghai is one of the busiest and most competitive
on the daily basis. The reason for a low security and privacy cities in the world and it has millions of different services on
scores is because all personal data and information have to offer. Interestingly a lot of IoT and IT can be classified into
be supplied for all transactions. If user data can be made Ali and non-Ali services as reflected by our findings.
anonymous and ways to provide real-time user authentication
can be made, this can enhance the level of security. It is per- VII. CONCLUSIONS AND FUTURE WORK
haps because in order to ensure a stable and fast connection, Finally, our proposed model can be used to estimate influ-
security and privacy tend to be regarded on a lower scale in ential factors of IoT-related enterprise. We aid decision mak-
these five representative providers in China. ers to identify the ideal solutions. Our proposed model can
Unfortunately, KNIME is not common in the UK and deal with vague, impression, and inconsistent information.
China. There are local solutions developed by service We enhance decision judgment by the use of AHP combined
providers. Due to this reason, they are classified under with neutrosophic sets. By using neutrosophic equations,

59572 VOLUME 7, 2019

N. A. Nabeeh et al.: Neutrosophic Multi-Criteria Decision Making Approach for IoT-Based Enterprises

the proposed alternatives have been chosen effectively using [14] B. Ma, C. Tan, Z.-Z. Jiang, and H. Deng, ‘‘Intuitionistic fuzzy multicriteria
neutrosophic rather than decision maker judgments. The con- group decision for evaluating and selecting information systems projects,’’
Inf. Technol. J., vol. 12, no. 13, pp. 2505–2511, 2013.
sistency rate approve that the use of neutrosophic sets will [15] V. Chang, M. Abdel-Basset, and M. Ramachandran, ‘‘Towards a reuse
enhance the inconsistent information that exist in decision strategic decision pattern framework—From theories to practices,’’ in
maker judgments matrix. We also replicated our proposal in Information Systems Frontiers. 2018. doi: 10.1007/s10796-018-9853-8.
[16] L. Wu, W. Cui, Y. Chen, and Y. Fu, ‘‘A group decision-making model for
the UK and China. We discussed results and explained the multi-criteria supplier selection in the presence of ordinal data,’’ in Proc.
rationale for getting different scores. Results show that our IEEE Conf. Service Oper. Logistics, Inform., 2008, pp. 1686–1690.
work can be adapted and replicated in different settings and [17] F. T. S. Chan and N. Kumar, ‘‘Global supplier development considering
countries for IoT research. Similarly, our findings for the risk factors using fuzzy extended AHP-based approach,’’ Omega, vol. 35,
no. 4, pp. 417–431, Aug. 2007.
smart city in UK and China were presented. [18] W. Khowfa and O. Silasai, ‘‘The integration of association rules and
The future work we are ongoing to predict the influential AHP in cloud service selection,’’ Int. J. Appl. Eng. Res., vol. 12, no. 24,
factors affecting enterprise by the use of variant multi-criteria pp. 15814–15820, 2017.
[19] M. Thirumaran and A. V. Arimathi, ‘‘Collaborative web service QoS
decision analysis methodologies, so that our research con- prediction with multi-criteria decision making using CB-NIMF,’’ Int. J.
tributions can be transferrable to other domains. In addition Comput. Sci. Inf. Technol., vol. 6, no. 2,pp. 1433–1438, 2015.
to, perform optimization of decision judgment matrices using [20] M. M. Akarte, N. Surendra, B. Ravi, and N. Rangaraj, ‘‘Web based casting
evolutionary algorithms. supplier evaluation using analytical hierarchy process,’’ J. Oper. Res. Soc.,
vol. 52, no. 5, pp. 511–522, May 2001.
[21] C. Muralidharan, N. Anantharaman, and S. G. Deshmukh, ‘‘A multi-
A. LIMITATION OF PROPOSED RESEARCH criteria group decision making model for supplier rating,’’ J. Supply Chain
More involvements from more companies will make our Manage., vol. 38, no. 4, pp. 22–33, Apr. 2002.
[22] F. T. S. Chan, ‘‘Interactive selection model for supplier selection process:
research better. An analytical hierarchy process approach,’’ Int. J. Prod. Res., vol. 4, no. 15,
B. COMPETING INTERESTS pp. 3549–3579, Aug. 2003.
[23] F. T. S. Chan and H. K. Chan, ‘‘Development of the supplier selection
The authors announce that there is no discrepancy of interests model—A case study in the advanced technology industry,’’ Proc. Inst.
concerning the publication of this research. Mech. Eng., B, J. Eng. Manuf., vol. 218, no. 12, pp. 1807–1824, Dec. 2004.
[24] F.-H. F. Liu and H. L. Hai, ‘‘The voting analytic hierarchy process method
REFERENCES for selecting supplier,’’ Int. J. Prod. Econ., vol. 97, no. 3, pp. 308–317,
[1] S.-E. Lee, M. Choi, and S. Kim, ‘‘How and what to study about IoT: Sep. 2005.
Research trends and future directions from the perspective of social sci- [25] F. T. S. Chan, H. K. Chan, R. W. L. Ip, and H. C. W. Lau, ‘‘A decision
ence,’’ Telecommun. Policy, vol. 41, no. 10, pp. 1056–1067, Nov. 2017. support system for supplier selection in the airline industry,’’ Proc. Inst.
[2] I. C. L. Ng and S. Y. L. Wakenshaw, ‘‘The Internet-of-Things: Review Mech. Eng., B, J. Eng. Manuf., vol. 221, no. 4, pp. 741–758, Apr. 2007.
and research directions,’’ Int. J. Res. Marketing, vol. 34, no. 1, pp. 3–21, [26] M. Abdel-Basset, G. Manogaran, A. Gamal, and F. Smarandache,
Mar. 2017. ‘‘A hybrid approach of neutrosophic sets and DEMATEL method for
[3] M. Abdel-Basset, G. Manogaran, M. Mohamed, and E. Rushdy, ‘‘Internet developing supplier selection criteria,’’ Des. Automat. Embedded Syst.,
of things in smart education environment: Supportive framework in the vol. 22, no. 3, pp. 257–278, Sep. 2018.
decision-making process,’’ Concurrency Comput., Pract. Exper., p. e4515, [27] J. Hou and D. Su, ‘‘EJB-MVC oriented supplier selection system for mass
2018. doi: 10.1002/cpe.4515. customization,’’ J. Manuf. Technol. Manage., vol. 18, no. 1, pp. 54–71,
[4] P. T. M. Ly, W.-H. Lai, C.-W. Hsu, and F.-Y. Shih, ‘‘Fuzzy AHP analysis 2007.
of Internet of Things (IoT) in enterprises,’’ Technol. Forecasting Social [28] B. Gaudenzi and A. Borghesi, ‘‘Managing risks in the supply chain using
Change, vol. 136, pp. 1–13, Nov. 2018. the AHP method,’’ Int. J. Logistics Manage., vol. 17,no. 1, pp. 114–136,
[5] F. Smarandache, ‘‘α-Discounting method for multi-criteria decision 2006.
making (α-D MCDM),’’ in Information Fusion. 2010, pp. 1–7. doi: [29] S. Wibowo and S. Grandhi, ‘‘Fuzzy multicriteria analysis for performance
10.13140/2.1.4832.7364. evaluation of Internet-of-Things-based supply chains,’’ Symmetry, vol. 10,
[6] T. L. Saaty, ‘‘How to make a decision: The analytic hierarchy process,’’ no. 11, p. 603, Nov. 2018.
Eur. J. Oper. Res., vol. 48, no. 1, pp. 9–26, Sep. 1990. [30] Y.-M. Wang and K.-S. Chin, ‘‘Fuzzy analytic hierarchy process: A log-
[7] M. Abdel-Basset, M. Mohamed, Y. Q. Zhou, and I. M. Hezam, ‘‘Multi- arithmic fuzzy preference programming methodology,’’ Int. J. Approx.
criteria group decision making based on neutrosophic analytic hierarchy Reasoning, vol. 52, no. 4, pp. 541–553, Jun. 2011.
process,’’ J. Intell. Fuzzy Syst., vol. 33, no. 6, pp. 4055–4066, 2017. [31] C.-T. Chen, C.-T. Lin, and S.-F. Huang, ‘‘A fuzzy approach for supplier
[8] M. Abdel-Basset, G. Manogaran, M. Mohamed, and F. Smarandache, evaluation and selection in supply chain management,’’ Int. J. Prod. Econ.,
‘‘A hybrid neutrosophic group ahp-topsis framework for quantifying risks vol. 102, no. 2, pp. 289–301, Aug. 2006.
in a supply chain,’’ Gener. Comput. Syst., vol. 10, no. 6, p. 226, 2018.
[32] A. Sarkar and P. K. J. Mohapatra, ‘‘Evaluation of supplier capability and
[9] M. Abdel-Basset, G. Manogaran, and M. Mohamed, ‘‘Internet of Things
performance: A method for supply base reduction,’’ J. Purchasing Supply
(IoT) and its impact on supply chain: A framework for building smart,
Manage., vol. 12, no. 3, pp. 148–163, May 2006.
secure and efficient systems,’’ Future Gener. Comput. Syst., vol. 86,
pp. 614–628, Sep. 2018. [33] R. Florez-Lopez, ‘‘Strategic supplier selection in the added-value perspec-
[10] S. Wibowo, H. Deng, and W. Xu, ‘‘Evaluation of cloud services: A fuzzy tive: A CI approach,’’ Inf. Sci., vol. 177, no. 5, pp. 1169–1179, Mar. 2007.
multi-criteria group decision making method,’’ J. Algorithms, vol. 9, no. 4, [34] W. Ho, X. Xu, and P. K. Dey, ‘‘Multi-criteria decision making approaches
p. 84, Dec. 2016. for supplier evaluation and selection: A literature review,’’ Eur. J. Oper.
[11] S. Wibowo and H. Deng, ‘‘Multi-criteria group decision making for eval- Res., vol. 202, pp. 16–24, Apr. 2010.
uating the performance of e-waste recycling programs under uncertainty,’’ [35] S. D. Pohekar and M. Ramachandran, ‘‘Application of multi-criteria deci-
Waste Manage., vol. 40, pp. 127–135, Jun. 2015. sion making to sustainable energy planning—A review,’’ Renew. Sustain.
[12] S. Wibowo and H. Deng, ‘‘Consensus-based decision support for mul- Energy Rev., vol. 8, no. 4, pp. 365–381, Aug. 2004.
ticriteria group decision making,’’ Comput. Ind. Eng., vol. 66, no. 4, [36] J. Sólnes, ‘‘Environmental quality indexing of large industrial development
pp. 625–633, Dec. 2013. alternatives using AHP,’’ Environ. Impact Assessment Rev., vol. 23, no. 3,
[13] C. H. Yeh, H. Deng, S. Wibowo, and Y. Xu, ‘‘Multicriteria group decision pp. 283–303, May 2013.
support for information systems project selection,’’ in Next-Generation [37] A. P. Athreya, B. DeBruhl, and P. Tague, ‘‘Designing for self-configuration
Applied Intelligence. Berlin, Germany: Springer, 2009, pp. 152–161. doi: and self-adaptation in the Internet of Things,’’ in Proc. 9th IEEE Int. Conf.
10.1007/978-3-642-02568-6_16. Collaborative Comput., Netw., Appl. Worksharing, Oct. 2013, pp. 585–592.

VOLUME 7, 2019 59573

 N. A. Nabeeh et al.: Neutrosophic Multi-Criteria Decision Making Approach for IoT-Based Enterprises

[38] M. Goyal, J. Lu, and G. Zhang, ‘‘Decision making in multi-issue e-market MOHAMED ABDEL-BASSET received the B.Sc.,
auction using fuzzy techniques and negotiable attitudes,’’ J. Theor. Appl. M.Sc., and Ph.D. degrees in information technol-
Electron. Commerce Res., vol. 3, no. 2, pp. 97–110, Aug. 2008. ogy from Zagazig University. His current research
[39] Y. Yamamoto et al., ‘‘IoT-aware online shopping system enhanced with interests are optimization, operations research,
gaze analysis,’’ in Proc. World Automat. Congr. (WAC), Jul./Oct. 2016, data mining, computational intelligence, applied
pp. 31–35. statistics and decision support systems, robust
[40] M. Abdel-Basset, M. Mohamed, F. Smarandache, and V. Chang, ‘‘Neutro- optimization, engineering optimization, multi-
sophic association rule mining algorithm for big data analysis,’’ Symmetry,
objective optimization, swarm intelligence, evolu-
vol. 10, no. 4, p. 106, 2018.
tionary algorithms, and artificial neural networks.
[41] M. Mohamed and F. Smarandache, ‘‘A hybrid neutrosophic group ANP-
Topsis framework for supplier selection problems,’’ Symmetry, vol. 10, He is working on the application of multi-objective
no. 6, p. 226, 2018. and robust meta-heuristic optimization techniques. He has published over
[42] M. Abdel-Basset, M. Mohamed, and V. Chang, ‘‘NMCDA: A framework 100 articles in international journals and conference proceedings. He is also
for evaluating cloud computing services,’’ Future Gener. Comput. Syst., an/a Editor/Reviewer in different international journals and conferences.
vol. 86, pp. 12–29, Sep. 2018.
[43] M. Abdel-Basset and M. Mohamed, ‘‘The role of single valued neu-
trosophic sets and rough sets in smart city: Imperfect and incomplete
information systems,’’ Measurement, vol. 124, pp. 47–55, Aug. 2018.
[44] M. Abdel-Basset, M. Gunasekaran, M. Mohamed, and N. Chilamkurti,
‘‘Three-way decisions based on neutrosophic sets and AHP-QFD frame- HAITHAM A. EL-GHAREEB is currently an
work for supplier selection problem,’’ Future Gener. Comput. Syst., vol. 89, Assistant Professor with the Information Systems
pp. 19–30, Dec. 2018.
Department, Faculty of Computers and Informa-
[45] A. G. Chofreh, F. A. Goni, and J. J. Klemeš, ‘‘Sustainable enterprise
resource planning systems implementation: A framework development,’’
tion Sciences, Mansoura University, Egypt. He is
J. Cleaner Prod., vol. 198, pp. 1345–1354, Oct. 2018. a member of many distinguished computer organi-
[46] A. Gunasekaran, K. Lai, and T. C. E. Cheng, ‘‘Responsive supply chain: zations, a Reviewer of different highly recognized
A competitive strategy in a networked economy,’’ Omega, vol. 36, no. 4, academic journals, a contributor to open source
pp. 549–564, Aug. 2008. projects, and the author of different books. He
[47] M. Bauer et al., ‘‘IoT reference model,’’ ARM Testimonials. doi: is interested in E-learning, enterprise architecture,
10.1007/978-3-642-40403-0_7. information architecture, especially in service ori-
[48] N. A. Nabeeh, F. Smarandache, M. Abdel-Basset, H. A. El-Ghareeb, ented architecture (SOA), business process management systems, virtualiza-
and A. Aboelfetouh, ‘‘An integrated neutrosophic-TOPSIS approach tion, big data, and in collaboration with the Information Systems E-learning
and its application to personnel selection: A new trend in brain pro- Organizations and Researchers.
cessing and analysis,’’ IEEE Access, vol. 7, pp. 29734–29744, 2019.
doi: 10.1109/ACCESS.2019.2899841.
[49] D. Ergu, G. Kou, Y. Peng, and Y. Shi. ‘‘A simple method to improve the
consistency ratio of the pair-wise comparison matrix in ANP,’’ Eur. J. Oper.
Res., vol. 213, no. 1, pp. 246–259, Aug. 2011.

NADA A. NABEEH received the B.S. and master’s AHMED ABOELFETOUH is currently a Professor
degree in information systems from the Faculty of of intelligent information systems, and also the
Computers and Information Sciences, Mansoura Vice Dean of Higher Studies with the Faculty of
University, Egypt. Her current research interests Computers and Information Sciences, Mansoura
include cloud computing, big data, smart city, University, Egypt. His research interests include
the Internet of Things, neural networks, artificial intelligent information systems, decision support
intelligence, web service composition, and evolu- systems, management information systems, and
tionary algorithms. geographic information systems.

59574 VOLUME 7, 2019