You are on page 1of 15

Cluster Comput

DOI 10.1007/s10586-017-0953-x

A study on fuzzy logic based cloud computing


Bashir Hayat1 · Kyong Hoon Kim1 · Ki-Il Kim2

Received: 30 March 2017 / Revised: 15 May 2017 / Accepted: 24 May 2017


© Springer Science+Business Media New York 2017

Abstract Cloud computing is now being deployed in real 1 Introduction


world to satisfy several users’ requirements for computa-
tion. In the point of management, there are several important Cloud computing has emerged as a paradigm to deliver
considerations such as availability, reliability, resource uti- requested services such as infrastructure as a service (IaaS),
lization, and throughput in cloud computing. However, since platform as a service (PaaS) and software as a service (SaaS)
these performance metrics are affected by the many uncor- over the network to the consumer under distributed comput-
related parameters, it is very hard task to derive new model ing architecture. The ambition of cloud computing is to earn
which takes into them account together. Even though there the maximum benefits from distributed resources by utilizing
are many feasible models, fuzzy logic can be the most suit- the resources utmost. To achieve this goal, cloud computing
able one in the view of depth, popularity and applicability merges the resources to attain higher throughput, minimum
in many other research areas. However, as far as the authors misuse of resources, and capable to knob large computation
know, there is only one short survey paper which focuses problems as explained in [1]. Thus, customer does not need to
on introducing research challenges without detail discussion worry about resources that might be needed in future. By this
of each mechanism. Based on this deficiency, in this paper, way, cloud computing has wiped hand over other computing
we present the state-of-the-art approaches and their impor- paradigms by improving the availability of resources greatly.
tant features in fuzzy logic based cloud computing. First, we For example, customer is able to use computing resource
present overview of cloud computing and categorization for without any manual interaction with service provider as well
the current research works. Second, we also provide some as whole resources that is intended to be uproot to the cloud
of the key techniques presented in the recent literature and for the users. Thus, users can rent out any resource dynami-
provide a summary of related research works. Finally, we cally without knowing the actual place of the resource lie [2].
suggest potential directions for future research in the field. The major challenge in cloud computing are virtualiza-
tion that is migrating of virtual machine (VM) and taking
Keywords Cloud computing · Fuzzy logic · Virtual intelligent decision over migrating time as critical challenge.
machine · Analytical model Selection of service and cloud path are another big issue in
cloud computing. For example, selection of optimal path to
a class of clouds that provide the same services with the rea-
sonable cost and selection of services in multiple selection
criteria should meet quality of service (QoS) requirements
of customer. Scheduling and load balancing are also known
B Ki-Il Kim to require reasonable solution in cloud computing. As for
kikim@cnu.ac.kr
allocation of resources and scheduling of task with limited
1 Department of Informatics, Gyeongsang National University, resources in cloud data center, it is demanded to balance load
Jinju 52828, Korea to control response time at proper level.
2 Department of Computer Science and Engineering, As mentioned before, there are many performance metrics
Chungnam National University, Daejeon 34134, Korea that should be concerned in cloud computing. Consequently,

123
Cluster Comput

since they are affected by various parameters which is not


correlated with each other, new model to evaluate impact
of respective parameters as well as integrate them in uni-
fied model is strongly required. For this goal, many existing
models can be applicable. Among them, fuzzy logic can be
the most feasible solution when it comes to depth, applica-
bility and low complexity. More detailed, fuzzy logic is a
problem solving methodology based on multivalued logic.
Fuzzy logic can model human mind through precise logi-
cal reasoning and provide reliable system between users and
cloud systems. Fuzzy logic looks similar to the way our brain
work. Some of the related essences of fuzzy logic are fuzzy
set, fuzzy rules, and linguistic [3]. Fuzzy set consist of collec-
tion of elements calculating unlikelihood in the set and has
various degree of membership varying from 0 to 1. Linguistic
variables represent values in words rather than number (e.g.
low, moderate, high). The rule base contains if then rules
which explain the operation of the controller.
According to proper applicability, many research works
employs the fuzzy logic in cloud computing to address men-
tioned research challenges. Even though there is one short
survey paper [4] which addresses the attributes of cloud
computing and research issue and provides example service
model for medical diagnostics using cloud computing with
fuzzy logic, it is not enough to provide collection of arti-
cles on a specific topic of fuzzy logic based cloud computing
Fig. 1 Layered structure of cloud computing
and attract high interest from the research community. To
defeat deficiency, in this paper, we present the state-of-the-art
approaches and their important features in fuzzy logic based Figure 1 visualizes the laminated structure of cloud. Vir-
cloud computing. We present the categorization and give an tualized resources are offered by IaaS on request. IaaS is the
overview of each scheme. Also, open issue is followed by lowest layer of cloud computing system as mentioned in [7].
summary and comparison of each category. Programmer can get platform for development of their own
The rest of the paper is organized as follow. In Sect. 2, we application at PaaS irrespective of how many memory and
describe the background of cloud and fuzzy logic. Research processor are needed [8]. SaaS mitigates the stress of soft-
work for fuzzy logic based cloud computing is presented ware maintenance of customers and ease development and
in Sects. 3, 4 and 5. Open issue will be given to future testing of provider [9,10].
research in following section. Finally, we make conclusion in In parallel with cloud computing, we address the overview
Sect. 7. of fuzzy logic. After being proposed by Zadeh in 1965, fuzzy
set theory, has been widely practiced to resolve the confusion
of human perception. It also efficiently resolves uncertainties
2 Fuzzy logic in cloud computing in possible information. Fuzzy logic is extensively applied
in application like image processing, data mining, network-
Cloud computing comes in the row of parallel, distributed ing and so on. General fuzzy controller consists of rule base,
and grid computing where information can be acquired from fuzzifier, inference system, and defuzzifier as shown in Fig. 2.
web browser by users according to their needs [5]. Varieties of While rule base contains set of rules which fuzzy control deci-
services are offered by cloud computing like storage, comput- sion is taken, fuzzifier converts numerical values into fuzzy
ing and communication infrastructure etc. Cloud computing values with the help of membership function. Inference sys-
services are categorized into three classes according to the tem applies the rules of the rule base and generates fuzzy
capability provided and service model [6]. conclusion. Fuzzy conclusion has been converted to single
numerical out with the help of defuzzifier. In distributed
1. Infrastructure as a service (IaaS) system, uncertainty exists at every step. Due to excellent
2. Platform as a service (PaaS) response of fuzzy logic to uncertainty and its logically rea-
3. Software as a service (SaaS) soning, fuzzy logic is widely applied in cloud computing.

123
Cluster Comput

Fig. 2 General structure of


fuzzy logic control systems

Based on mentioned properties, we categorize the existing membership function acquires lot of advantages due to its
fuzzy logic based cloud computing into three parts accord- smoothness [20]. The mathematical is expression is Gaus-
ing to its type of membership function. Since membership sian membership function is
function defines how each point in the input space is mapped
to a membership value between 0 and 1, a specific curve −(x−C)2
f (x, α, C) = e 2α 2 (1)
is generally derived. According to shapes of membership
function, fuzzy inference system may be affected. So, usu-
where the parameter C is at the distance from center, α repre-
ally they have different shapes such as Gaussian, triangular,
sent the width of the curve. The use of Gaussian membership
and trapezoidal. So, we choose this membership function
function by fuzzy logic in cloud computing are given below.
as the first level categorization. At each category, we divide
existing schemes into four different groups according to the
combination method. Instead the research work for cloud 3.1 Non heuristic model based methods
computing is not completed by fuzzy logic; they combine
fuzzy logic with other model to achieve goal which includes Zulkar Nine et al. [21] modeled dynamic load balancing
resource optimization, load balancing, scheduling and oth- algorithm based on fuzzy logic in cloud computing. When
ers. The good example of cooperative schemes include first new job arrives to Data Center (DC) controller, fuzzy active
come first service (FCFS), Different Evaluation, and analytic monitoring load balancer (FAM-LB) brings suitable virtual
hierarchy process (AHP) and so on. The brief introduction machine for the required resource of the request based on the
for each model is summarized in Table 1. memory, disk space and bandwidth stature. Based on mem-
According to their property, we make four groups, ory, bandwidth and disk space, FAM-LB revises the virtual
that is, non-heuristic method, heuristic method, multi- machine where next task is to be scheduled. The proposed
attribute decision model (MADM)/multi-criteria decision model makes prediction which VM will receive the coming
model(MCDM), and hierarchical method. As a final, cate- job but it did not calculate the requirement for the request to
gorization of fuzzy based cloud computing is illustrated in achieve high level of trust.
Fig. 3. From now, we explain their key objective and features. Dynamic resource scheduling and system accessibility
play a vital role to improve the quality of cloud comput-
ing environment. The proposed method covers the limitation
3 Gaussian membership function based cloud up to some extent by calculating the relationship between
computing resources and the requirements for resource. More detailed,
a novel method called moving average based fuzzy resource
Gaussian membership function is established by consider- scheduling (MV-FRS) was suggested to optimize resources
ing Gaussian function, developed by well know scientist scheduling in virtualized environment while considering
Carl Friedrich Gauss. Gauss function is harmony of expo- bandwidth, memory and processing cycle. The goal is to
nential and quadratic function opening downward. Fuzzy reduce the total waiting time and enhance resource schedul-
model along the adoption of Gauss function are nonlinear. ing efficiency. Reduction of total waiting time is achieved
The Gauss functions ease to achieve glossy, deliberately dif- by estimating the resources with the help of moving aver-
ferentiable hyper surface of Fuzzy model [58]. age method. In addition, the MV-FRS enhances the normal
Gaussian membership function as shown in the Fig. 4 achievement rate of the resources being assigned by resource
is used to represent uncertain, linguistic terms. Gaussian manager through employing Markov Decision method.

123
Cluster Comput

Table 1 Summary of methods used with fuzzy logic in cloud computing


Method Explanation

First come first serve (FCFS) It is mentioned as first in first severe. It is policy that entertains the requester in the order
they arrived. The entire requests which are waiting for their turn for processing are in
queue. This policy is extensively employed for scheduling
Randomize round robin It is the improved version of round robin by selecting the process randomly in case
different tasks come
Conventional round robin It assigns fixed time slice to each process in a circular fashion where performance is
evaluated according to the size of the time slice
Differential evolution (DE) [11] DE gets optimization of problem by iterative procedure. In DE, few or no supposition
are considered initially to optimize the problem.
Particle swarm optimization (PSO) [12,13] Particle Swarm Optimization optimizes the problem over a few number of iteration.
Group of variables adjusts their values to the variable whose updated value is closest to
the target value in any iteration
Artificial bee colony/honey bee colony (ABC/HBC) ABC is the latest algorithm developed according to the behavior of honey bees swarms
[14]
Technique for order of preference by similarity to TOPSIS is based on the selection of alternative that has less distance from Positive Ideal
ideal solution (TOPSIS) [15] Solution (PIS). In TOPSIS each Criteria has been evaluated against certain criteria
Simple additive weighting (SAW) [16] SAW is a one of multi-attribute decision model (MADM) technique based on the
weighted average. Evaluation score is calculated for every alternative by multiplying
scaled value that is assigned to the alternative followed by adding the product of all
criteria
Analytic hierarchy process (AHP) [17] AHP is effective tool to deal with the complexity since it reduces complexity with
iterative pairwise comparison. Initially AHP divides the problem into hierarchy, each
of which will be analyzed separately
VIseKriterijumska Optimizacija I Kompromisno It is one of the most applicable multi-attribute decision model developed by Serafim
Resenje (VIKOR) [18] Opricovic to resolve problem having conflicting and different standards for
measurement. VIKOR mark the alternate and conclude the that it is the nearest ideal
Gray rational analysis (GRA) [19] GRA has the ability to solve the problem having complicated interrelationship among
multiple attributes. GRA unifies the total available performance attributes values and
consider for each alternate into single value

Moreover, Priya et al. [22] design Max Min Fuzzy infer- utilization and size, to map resources for cloud services.
ence to improve resource scheduling as discussed. In the Initially the tasks are classified based on the above men-
proposed model, they ignore qualitative attributes to mea- tion criterion. Neural network will take decision to assign
sure QoS of the suggested scheme further. Prediction and the system resources for the upcoming tasks based on con-
optimization occur collectively whichever there is need to verted input data from the fuzzification. Genetic algorithm
create confusion in certain cases. is employed to align the user requests with the available sys-
In order to improve the availability and response of the vir- tem resources in fuzzy neural network. The proposed model
tualized DC, an online dynamic scheduling algorithm [23] is outperforms in smaller system as compared to the large
proposed to build excellent tradeoff between major objec- one.
tives, that is, load balancing and prediction. The proposed
scheme can receive multi-class tasks characterized by their 3.2 Multicriteria decision model based methods
specific arrival rates, service time and availability require-
ments. Availability fuzzy predictor calculates the availability Srimannarayana Gran et al. [25] map the performance assess-
of virtual execution unit (VEU), by using hardware, virtual ment of cloud service provider as a multi trait cooperative
machine monitor (VMM), and VM in Type I fuzzy logic judgment making problems. And, it shows a fuzzy multi
system. Load balance fuzzy predictor forecasts the load sit- attribute collective decision making model for efficiently
uation using auto regression, and type II fuzzy logic system handling the performance assessment issues of cloud server’
considering processing capacity and workload as parameters providers. The membership function in interval rather than
of each VEU. single value to avoid vagueness reduces cognitive demand on
Venkatesa et al. [24] introduce new mechanism using the decision making process. Interval-valued fuzzy numbers
neural network with fuzzy logic to observe the impact of are applied on four criteria cost, that is, scalability, adaptabil-
the same input as the previous work like bandwidth, CPU ity, service response time and suitability. Distance between

123
Cluster Comput

Fig. 3 Categorization of research work for fuzzy logic based cloud computing

positive ideal and negative ideal is computed through Haus- scheme employed neuro-fuzzy for prediction the resource
dorff metric. usage, whether the resources are underutilized or overloaded
Mya and Thein [61] suggest a scheme to deal with Vir- as neuro- fuzzy have the capability to reasoning and learning
tual machine load prediction and migration. The suggested in vague situation. Checking the consistency of the weights of

123
Cluster Comput

along complexity of virtualized DC and dynamically chang-


ing loads, fuzzy logic is used in the controller. The local
controller is attempting to cut down the resource consump-
tion to minimize the resource cost fulfilling service level
agreement (SLA) of its client. On the other hand, the global
controller tries to enlarge its own profit. Workload, applica-
tion performance and resource usage of the virtual container
are the building block of fuzzy modeling and inference func-
tions in local resource controller.
Niyati et al. [27] Presented gaussian membership func-
Fig. 4 Gaussian membership function
tion based Hierarchical fuzzy logic to estimate the services
in cloud. Considering fourteen attributes hierarchical fuzzy
system scheme called hierarchical fuzzy system semantic
criteria is done by AHP whereas Fuzzy TOPSIS is applying web application as a service (HFS-SWAaaS) for quality mea-
for alternate ranking. surement of semantic web based application delivered as
service on cloud. The proposed scheme knobs the multiplic-
ity of quality attributes. It also holds unique quality attributes
3.3 Hierarchical model based methods
of Semantic Web Applications(SWAs) which is inferable
from its service and in addition semantics. The output values
Two level resource management system is proposed with
obtained from low level fuzzy system are considered as input
local and global controller at virtual container level and
to the next level of system.
resource pool level, respectively. The proposed hierarchical
In summary, we illustrates thee comparison of the method
fuzzy logic system [26] consists of following parameters, per-
cooperating with fuzzy logic with Gaussian membership
formance efficiency, cost of communication between VMs
function in Table 2.
and total power. To handle the uncertainty and robustness,

Table 2 Gaussian membership function based fuzzy cloud computing


Ref. Cloud service Method Decision parameter Correlation Objective Technique
layer

[21] Iaas Non Bandwidth, disk Round robin LB, Predict efficient virtual Fuzzy active
heuristic space, memory min-min LB, active machine monitoring load
monitor LB balancer (FAM-LB)
[22] Cloud Non Resource requirements Collaborative cloud Reducing total waiting Moving average fuzzy
computing heuristic (bandwidth, computing time, improving resource scheduling
memory, CPU) harmony(CCC-H), resource scheduling (MA-FRS)
hollaborative load efficiency
management protocol
(CProtocol)
[23] IaaS Non Hardware Min-Min scheduling Performance and Scheduling algorithm
heuristic components, virtual and suffrage availability based on load balance
machine monitor, scheduling algorithm requirements of SLA and availability fuzzy
virtual machine, (ART, Satisfaction prediction (SALAF)
CPU process queue Percentage)
length, memory
[24] Cloud Non Bandwidth, memory Berger model Improve completion time Fuzzy neural network
computing heuristic and expected time of task, efficient
bandwidth utilization
[25] – MCDM – Interval-valued Performance evaluation Fuzzy multi attribute
intuitionist fuzzy of cloud computing group decision
numbers provider making algorithm
[26] – Heuristic Number of dynamic Ideal situation High quality of service Two level resource
request with lower resource management system
allocation cost using fuzzy modeling
(throughput achieve)

123
Cluster Comput

cost. Dorian and Bernd [31] proposed a solution to support


VM in assigning resources operating on physical machine
in the cloud environment. To support scalability for huge
amount of VMs, multi agent fuzzy controller is introduced
for assigning resources to a distinct VM. To observe perfor-
mance cost for VM in cloud, experimental comparison multi
agent fuzzy controller, fuzzy controller and state of the art
adaptive optimal control approach are considered. But, they
fail to calculate whether the allocated resources fulfill the
requirements of the user and ignore nonoperational parame-
Fig. 5 Triangular membership function
ters in the system.
Zhijia et al. [32] presented a scheme that has the ability
4 Triangular membership function based cloud to consider attributes which are not commonly taken into
computing account like system friendliness. The suggested dynamic
resource scheduling scheme established on fuzzy theory sat-
Straight line functions are generally attractive through its isfies numerous consumers’ demands and educated system
candor. Triangular function comes in this category. Fewer to predict consumers’ fuzzy requests. The suggested scheme
amounts of data are desired to define the membership func- assigns appropriate resources to the user’s requests using
tion. Triangular membership function may be characterized fuzzy control model. Before assigning resources to the user’s
as threefold numbers [59]. A triangular membership function requests, estimation of resources has been determined in the
is shown in Fig. 5 and is described by three parameters α, β prediction layer which followed by completing calculation
and γ , given in the mathematical expression. of resource availability and resource requirements. Fluctu-
ating criterion is sent through feedback control to optimize
   
x −α γ −x resource allocation model. FCFS queue is ratifying to man-
f (x, α, β, γ ) = max min , ,0 (2)
β −α γ −β age multiuser tasks. Fuzzy control takes decision of scale up
or scale down on the basis of tasks requirements for resources
The values of α and γ , locates at the minimum and whereas and available resources.
β does the peak position of the triangle [28]. Fuzzy logic
model using triangular membership function in cloud com- 4.2 Heuristic model based methods
puting are given.
In [33], Luca et al. studied the issues of cloud application
4.1 Non heuristic model based methods managementand suggest fuzzy controller for allocation of
VM correlated to the cloud application. Its goal is to achieve
In [29], Srimannarayana et al., suggested fuzzy based round minimum amount of physical capacity desired to accommo-
robin load balanced algorithm in VM of cloud computing. date application service level objective. Proposed controller
Fuzzification process considers processor speed and load on shows improvement in quality of service and in minimizing
VM as input parameter in order to generate balanced load. energy consumption. RUBiS application is controlled with
This is further needed for inference system. The efficiency is the proposed fuzzy Q&E model in the real test bed and com-
improved by minimizing the processing time of data center pared with DynaQoS to shows better results. Considering
(DC). Moreover, response time is reduced as compared to memory and disk space along with different application, the
conventional round robin algorithm. performance may vary with the suggested model.
To perform better scheduling for VM, Amin et al. [30] Heuristic Fuzzy logic based triangular membership func-
present fuzzy based scheduling algorithm. The fuzzy con- tion is applied to optimize the cost of renewable energy
troller enforces two variables as input i.e., number of request consumption in cloud DC. The proposed model diminishes
in host and consider a task having highest order from each the cost without prior to judgment of forthcoming elec-
VM. The request number is named as AGE whereas task tricity rate, availability of renewable energy and upcoming
Precedence as PREC. VM id is produced by fuzzy controller workload. Average price of electricity, usage of renewable
which will be assigned to host. It reveals better results rather energy and volume of brown energy consumption in DC
than RR and FCFS in term of time (reduce in completion). within a certain time window of recent past are considers
However, the authors ignore the drastically changed request as input by fuzzy inference engine. The upcoming requests
and the impact of proposed scheme. To overcome the above are handled with FCFS model. Adel and Rajkumar Buyya
issues, dynamic resource allocation for VM in cloud is crit- [34], presented Mamdani fuzzyinference system with cen-
ical to ensure provision execution and decrease operating troid of area defuzzification having triangular member ship

123
Cluster Comput

function. Fuzzy logic based load balancing (FLB) figures is engaged to capture user’s abstract requirements and desire
out the appropriateness of all DC’s and routes the new to disparate QoS criteria’s followed by hierarchical fuzzy
request to the highest degree of suitable DC. Fatimeh et inference system to delivered trust degree.
al. [35] apply fuzzy particle swarm optimization (FPSO) to Seyedeh et al. [39] apply hierarchal fuzzy using triangu-
remove inured obstruction of VM mapping in cloud com- lar membership to handle the complexity of VM migration.
puting and minimize the power consumption. The suggested Power consumption, workload of the nodes, temperature, and
FPSO is intended to identify optimum PMs over which VM availability have a great impact on ranking the promising
are mapped, and improve resource utilization by minimizing physical machine for a migrating VM. Hierarchical fuzzy
power consumption and reducing VM transfer. logic system generally works in two tiers. calculating PM
Christian et al. [36] investigate the issues of service selec- rank on the basis of serving conditions, computation of PM
tion for storage in cloud computing, in particular, fuzzy rank on the basis of communication cost, and computation
concept is adopted to stem the subjective imprecise prefer- of PM rank based on power consumption act as three facet
ences of user for the sought QoS level. The selection issue is of First tier whereas fuzzy decision controller is functioned
cleared up with fuzzy inference and Dempster Shafer theory in the later tier.
to build a selection about the finest service matching the user
requirements. Distributed formulas are suggested in order to 4.4 Muliticriteria model based methods
get optimum results in a more efficient way. Vickrey, Clarke
and Groves (VCG) is applied on the solution to endorse the To evaluate the QoS of Wb Services, Chi et al. [40] presented
truth telling ones among the active service provider. The pro- fuzzy TOPSIS that helps service consumers’ and providers
posed scheme gives attention to one user at a time and grants to figure out available web services. The weights of crite-
only one service to fulfill the requirements. ria is represented by linguistics variables of triangular fuzzy
Simple additive weighting (SAW) scheme Saripalli et al. number, which can be molded to crips numbers using graded
to figure out and give preference to the cloud computing ser- mean integration representation method. Minkowski distance
vices in an organization was described in [37]. Cloud decision function is used to find out positive ideal solution (PIS) and
maker (DM) assigns points to each cloud computing service negative ideal solution (NIS) of each alternative web service.
and aggregated score decision for selection occurs among As the number of user increases the proposed fuzzy TOP-
the existing cloud computing services. Multiple attribute SIS method is efficient opposite to Huang’s work, Similarity
decision methodology for adoption of clouds (MADMAC) aggregation method (SAM). Huaming et al. in [41] proposed
conducts for each decision area (DA), respectively. It needs AHP and TOPSIS to find the optimum path for cloud offload-
the definition of attributes, attributes importance, and alter- ing, AHP is combined with fuzzy TOPSIS. Six criteria are
natives to develop decision matrix. And, it arrives at a relative considered for cloud path selection. AHP and Fuzzy TOPSIS
positioning to establish the optimal alternative. Attributes are working together provide good result for selecting most suit-
considered in a hierarchy fashion by having six main crite- able path while considering multiple criteria simultaneously.
ria that is subdivided into twenty seven. Wide band Delphi The decision hierarchies are composed on three layers: target
method is suggested to determine (a) relative weight of each layer, criteria layer, and decision layer as described.
attribute, (b) rate of cost/benefit for each attribute. Similarly, an extensive variety of criteria is utilized to
Boominathan and Aramudhan [62] apply firefly colony in evaluate the nature of distributed computing administrations.
data center. The objective is to minimized physical machine These vast majority criteria have associated or intelligent
in use in the form of server consolidation and virtual machine attributes which can make it hard to successfully break down
placement to decrease the wastage of resources. Reducing and increment utilization goals of cloud computing admin-
physical machine aim is to minimize the power consumption istrations. Hybrid fuzzy MCDM is adopted that incorporate
of data center, as compared to Ant colony, max-min Ant fuzzy DEMATEL, Fuzzy DANP and Fuzzy VIKOR method
system and first fit decreasing heuristic the Suggested scheme to enhance benefit level and accomplish the objective of
out perform. meeting user level requirements in fuzzy environment. Deci-
sion making trial and evaluation laboratory (DEMATEL)
4.3 Hierarchical model based methods [42] accomplish to construct a fuzzy scope influential net-
work relationship map (FSINRM), fuzzy DANP conclude the
Chenhao and Rajkumar [38] adopted fuzzy interface system influence weights of the criteria, followed by fuzzy VIKOR
to evaluate the trust of cloud services according to user’s that perform calculation, to figure out the difference of the
fuzzy demands and service vigorous performance to expe- desired level of service delivery. The main focus on study
dite service selection. To scope the appropriateness of cloud from the provider side is to consider the consumer side QoS
as their acceptable degree to a particular user’s demands is parameters clearly. Hamzeh et al. [43] studied the problem
based on their preceding performance. Membership function of evaluating and selecting cloud services according to trust

123
Cluster Comput

criteria, hybrid fuzzy modified VIKOR method is applied to clarity and computational effectiveness trapezoidal member-
achieve minimum cost and high performance. Fuzzy modi- ship function is widely used in image processing, distributed
fied VIKOR method has the capability to solve multi-criteria system etc. the function consist of straight, due to which
decision entanglements including contradictory and contrast- trapezoidal function have lack of smooth corners [60]. As
ing criteria. Such procedure can help cloud service requester we discussed from mathematical point of view, we can get
(CSR) to choose which cloud service provider (CSP) can different shapes of membership function. Trapezoidal mem-
meet their trust necessities before handing-off application, bership function depends on four parameters α, β, γ and θ
files and information to cloud. Nine step sequential proce- [28]. Mathematical expression is given below.
dures is designed with hybrid fuzzy and modified VIKOR    
method based on five decision maker having fifteen criteria. x −α θ−x
f (x, α, β, γ , θ ) = max min , 1, , 0 (3)
When it comes to consider consumer based QoS criteria, β −α θ −γ
Masoumeh et al. [44] presented a new framework to solve the
service selection problem of cloud services, based on trian- α and θ locate at the lowest position whereas β and γ do at
gular fuzzy logic. For more reliability, third party put-upon the shoulders. Figure 6 graphically shows trapezoidal mem-
to gather the data from published information of cloud ser- bership function. In the below section, we discuss detail use
vice providers, user feedback, monitoring tool information of trapezoidal membership function based fuzzy logic with
and user claimed priority. Fuzzy control module converts lin- existing methods in cloud computing in different applica-
guistic weights of parameters to the triangular fuzzy numbers tions.
and QoS modeling layer designed the most valuable criteria,
their liaison, and measuring the QoS needs to usable service 5.1 Non heuristic model based methods
configuration. To avoid exaggerated data of service capa-
bility published by cloud service provider on cloud service To enable load efficient data center in cloud computing,
repository is designed while containing data of cloud ser- Jaiganesh and Vincent in [47] presented fuzzy expert sys-
vice provider in standard format. Chinu and Sakshi in [45] tem. Three main stimulation of DC are act decision variables,
apply AHP for Service selection, that only analyzes, ranks namely, bandwidth, memory and CPU frequency, which
and selects the service to fulfill the requirements of end user. result in load efficient DC. The variables of the analysis are
Fuzzy analytic hierarchy process (FAHP) comes in a class of calibrated into three tiers. In the proposed scheme, there is
multi decision making approach. To contend uncertainty and no such module to equalize the incoming request from the
complexity issue in cloud path selection, FAHP is employed consumer. Different types of resource are created to cate-
to get optimal results considering five attributes as criteria. gorize cloud resources at the service level of user. Wang
Among them, speed and availability criteria play a vital role et al. [48] designed Cost based fuzzy clustering algorithm
to attain less execution time and minimum overhead. by accompanying four scheduling algorithms to analyze the
Tarighi et al. [46] investigate the use of MCDM in service jobs submitted by the user. In fuzzy clustering model (FCM),
selection of cloud computing, Fuzzy TOPSIS is also applied weight factor is introduced to scale the interval between class
in hierarchical fashion to migrate VM between cluster nodes. center and resource point. Resource clustering depends on
Three types of programs are operated simultaneously in the User Service Level (USL). And, its service categorization
scheme. Supervisor program has the responsibility to observe is in gold, silver and copper. Dispatching resource (DR),
usage of physical servers as well as find overloaded nodes. At schedules parallel jobs using FCFS, small job first first served
the time of predicament level, two programs will be activated (SJFS), large job first served (LJFS), and adoptive first come
to migrate VM automatically. The decision is taken on the first served (AFCFS). Performance is ascertained consid-
basis of CPU usage, network utilization, RAM usage and ering average response time (ART), average waiting time
multiple resources of nodes heavily utilized. (AWT), and Average running time of algorithm.
Table 3 shows the comparison of the methods used with the Te jen Su et al. [49] considered trapezoidal member-
fuzzy logic in different domain of cloud computing aiming ship function in fuzzy logic to adjust the working hours
to different objectives. of electrical appliances automatically. Fuzzy based power
monitoring system is designed in cloud environment. Power
monitoring module predicts the voltage and current of indi-
5 Trapezoidal membership function based cloud vidual electrical device and sends the data to host server
computing over Ethernet. Sever contains monitoring and control inter-
face that uses fuzzy logic to observed the on-off state of the
As previously mention straight line function are usually con- electrical devices. Using min-max fuzzy interface rule and
sidered because of its simplicity and ease to use. Trapezoidal weighted average for de-fuzzification taking humidity and
function also comes in this category as triangular. Due to its temperature, the system proposed generates working time of

123
Table 3 Triangular membership function based fuzzy cloud computing
Ref. Cloud service Method Decision parameter Correlation Objective Technique
layer

123
[21] IaaS Non heuristic Bandwidth, disk space and Round Robin LB, Min-Min Predict efficient virtual machine Fuzzy active monitoring load
memory LB, Active monitor LB balancer (FAM-LB)
[23] IaaS Non heuristic Hardware components, virtual Min-Min scheduling and Performance and availability Scheduling algorithm based on
machine monitor, virtual suffrage scheduling requirements of SLA (ART, load balance and availability
machine, CPU process queue algorithm Satisfaction Percentage) fuzzy prediction (SALAF)
length, memory
[29] IaaS Non heuristic Processing speed and virtual Conventional Round Robin Minimize processing and response Fuzzy based Round Robin (FRR)
machine load time
[30] IaaS Non heuristic No. of VM request sent to the host First come first serve Round Virtual machine Id assign to host Fuzzy scheduling
(AGE), and task precedence run Robin (suitable VM)
on VM (PREC)
[31] Cloud Non heuristic Value of change in utility and Centralized fuzzy controller Optimized utility function Multi agent fuzzy controller
change in resource allocation (FC), adaptive optimal (performance, cost)
control (OPC)
[36] Data center Heuristic Availability, reliability and latency, Dempster shafer theory and To store data in DC by providing best Fuzzy set theory with
security fuzzy interface storage service provider in Dempster-Shafer and TOPSIS
uncertainty, fulfilling the customer
requirement
[37] – Heuristic Benefit, cost, and non-monotonic – Improve cloud adaptation involving multiple attribute decision
attributes. multiple conflicting criteria methodology for adoption of
clouds (MADMAC)
[39] – Hierarchical Workload, availability, AppAware Algorithm VM migration to the appropriate PM Hierarchical fuzzy logic structure
communication cost, power with reducing communication cost for ranking potential destination
consumption(8) and improving performance of PM(HFLSRPM)
efficiency
[41] – MCDM Band width, price, speed, security, – Selection of an optimal cloud path in AHP and Fuzzy TOPSIS
availability cloud offloading systems
[42] – MCDM New business, reduce time – Adaptation of cloud service avoiding Hybrid Fuzzy MCDM
performance, security. (12) interdependence and feed back
problems
[43] – MCDM Data ownership, Security, – To minimize the trust uncertainty Fuzzy modified VIKOR
sustainability(15) between cloud service provider
(CSR) and Cloud service requester
(CSR)
[45] – MCDM Speed, availability, privacy, – Selection of cloud for offloading data Fuzzy Analytic Hierarchy Process
Capacity, cost with less execution time and (FAHP)
minimum overhead
[46] – MCDM CPU, RAM usage QoS(5) – Relocation of overloaded VM Hierarchical Fuzzy TOPSIS
[49] Cloud Non heuristic Pro speed, que length, pro power, Randomize Round Robin, Average response time and through put Two level fuzzy algorithm (Fuzzy)
computing Avg que length Round Robin
Cluster Comput
Cluster Comput

ing minimum migration time (MMT), maximum correlation


(MC) and steady resource consuming VMs as input by fuzzy
logic system in the suggested scheme.

5.3 Muliticriteria decision model based methods

To enhance the resource evaluation, Boom et al. [54] intro-


duced fuzzy logic for the dynamic service placement and
Fig. 6 Trapezoidal membership function
replication (DSPR). DSPR is self-management framework
that performs as the middle-ware for cloud computing
resources and use. The end goal is to permit DSPR to reg-
electrical devices. Smart phones and other communicating
ulate high level instruction through fuzzy logic. Two steps
devices can schedule electrical devices remotely.
procedure defines the policies and preference using FIS. Ini-
tially, range is set to the input variable, followed by level of
5.2 Heuristic model based methods acceptance that is characterized by way of production rule of
FIS. Fuzzy logic based DSPR enhances service availability
HBC with the use of fuzziness is applied to access different at the occurrence of multiple continuous deterioration.
services from cloud infrastructure. In suggested framework,
Hadoop framework is placed on the top and having four main
components which have different specific job user interface, 5.4 Hierarchical model based methods
authentication check, computational web service mapping,
and job scheduler. It can be accessed by light weight comput- To handle the problem of power consumption, Pooyan et al.
ers, mainframes, supercomputers and so on. Idle commodity [55] present new electricity controller, named RobusT2Scale,
systems are used to optimize and enhance the cost/benefit addresses the issue of parameter’s value prediction time,
analysis of cloud computing. Fuzzy Bee Colony in [50] is qualitative specification of thresholds, and robust control of
applied to compute the difference among different type of ser- uncertainty. Intention of the controlled elasticity is to raise
vices and satisfaction of user demands. The proposed model and reduce resources to meet the SLO. The proposed scheme
improves performance and reduces overhead over the sys- is vigorous to noisy data, which are collected based on user
tems as presented. side application level measurements. Fuzzy sets and rules
While considering the application of heuristic based fuzzy of the controller are accountable for managing the elasticity
logic to balance load in DC. Uma and Sangeev [51] presented reasoning. Rules in the elasticity reasoning receive multiple
glowworm swarm optimization (GSO). The model is to bal- input to construct output. Trapezoidal membership function
ance load onto different VMs of DC in public cloud in terms is created for workload and response time. Results are col-
of cost and time. Fuzzy logic and global swarm optimization lected by introducing PaaS services of the cloud.
are deploy in a switch mechanism manner to balance load in Trapezoidal membership function based on hierarchical
VM. Fuzzy logic is operated during idle state where as GSO method is applicable to minimize the power consumption.
does in the normal status. Load of VM and processor speed Mohammad Alaul et al. [56], consider minimum migration
are taken to produce better results to equalize the load of vir- time, correlation and migration control metric for steady
tual machine. The proposed scheme does not show whether resources consuming VM. It takes input variable for the
the need of resources is changed after assigning. Similarly, model using fuzzy logic along with VM selection. The pro-
Fahimeh et al. [52] studied resource handling in IaaS Cloud. posed model provides acceptable result in term of minimum
The proposed scheme is divided into have two main modules, power consumption and less SLA violation as compared to
resource monitor and decision making. To handle uncertainty the existing methods. Muhammad ali et al. [57] present two
in the resources, fuzzy Bayesian network puts into action. set fuzzy approaches are suggested. In first step, fuzzy is used
Mohammad Alaul et al. [53] take the energy consumption to figure out level of the load at cluster while considering pro-
issue into account in parallel with quality of service during cessing speed and queue length of the nodes. Moreover, in
VM migration from one host to another. Fuzzy logic based the next step, second level fuzzy approach is used for load
VM consolidation method is employed to establish harmony equilibrium at cloud by taking processing speed and average
of energy and QoS. The said scheme is based on three types queue length of cluster.
of algorithm, (1) core flow of virtual machine consolidation, The detail comparisons of the methods with fuzzy logic
(2) identification of less loaded VMs, and (3) overloaded having Trapezoidal membership function are given in the
detection algorithm to identify highly loaded hosts. Consider- Table 4.

123
123
Table 4 Trapezoidal membership function based fuzzy cloud computing
Ref. Cloud service layer Method Decision parameter Correlation Objective Technique

[36] Data center Heuristic Availability, reliability and Dempster-shafer theory and fuzzy To store data in DC by providing Fuzzy Set theory with
latency, security interface best storage service provider in Dempster-Shafer and TOPSIS
uncertainty, fulfilling the
customer Requirements
[47] Data center Non heuristic Bandwidth, memory, CPU HPL (LINKPACK Scheme) Optimization of data center load Data Center Load efficiency (DCLE)
cycles(Utilization)
[50] – Heuristic – Web service with one processor, Single objective (reduce Honey bee colony with fuzziness
one processor with agent, and overhead)
entirely four processors
[51] Public cloud Heuristic VM load and processing speed Round Robin Minimize processing and Fuzzy and GSO (Fuzzy with swarm
response time in term of cost optimization)
and time
[53] Data center Heuristic Minimum migration time, Dempster-shafer theory and Fuzzy To reduce energy consumption Fuzzy Virtual Machine Selection
correlation, migration control interface and minimum SLA violation Method (MMSD_FS)
matrix
[54] – MCDM Availability, cost, performance – Selection of best machines from Dynamic service placement and
resource pool to meet the Replication (DSPR)
requirements of service
[56] – Hierarchical Minimum migration time, Combination of 5 Built in overload VM selection and migration Fuzzy VM selection with migration
correlation, migration control detection algorithms and three control to minimize the power control
metric for steady resource built in VM selection algorithm consumption with minimum
consuming WM (Cloud Sim) SLA violation
[57] Cloud Hierarchical Pro speed, queue length, pro Randomize Round Robin, Round Average response time and Two level fuzzy algorithm (Fuzzy)
power, average queue length Robin through put
Cluster Comput
Cluster Comput

6 Research challenges ods. We compare several methods by reviewing the exist


literature. Different kind of domains have been selected the
Even though respectable amount of research has been con- use of fuzzy logic in cloud computing and discussed in the
ducted in fuzzy based cloud computing by addressing the points of advantage and limitation. Moreover, different meth-
significance of fuzzy and suggesting new solution to the prob- ods, heuristic, non-heuristic, MCDM, and hierarchical which
lem in cloud computing, we still have some issues that need usually operates together, are compared. In the end we dis-
to be address in further studies. cuss some open issues and challenges which will aid new
researchers in selecting research directions.
• Virtualization is to enhance the utilization of resources
and minimize the operation cost in DCs of clouds. Dif- Acknowledgements This research was supported by Basic Science
Research Program through the National Research Foundation of
ferent type of input parameters are taken to achieve the Korea (NRF) funded by the Ministry of Education (Grant no. NRF-
above mention objective, but we need to consider the 2015R1D1A3A01019680) and “Human Resources Program in Energy
complete application running on the higher layer of VM Technology” of the Korea Institute of Energy Technology Evaluation
for choosing the appropriate physical machine for host- and Planning (KETEP), granted financial resource from the Ministry of
Trade, Industry & Energy, Republic of Korea (No. 20174030201440).
ing those VM. It allows the possibility to consider the
communication reliance between VM in multiple level
enterprise utilization and the fundamental DC network
topology and maximum limits of physical machine in References
the DC.
1. Wahiduzaman, M., Gani, A., Anur, N., Shiraz, M., Haque, M.,
• Various fuzzy VM migration methods exist for migrat- Haque, I.: Cloud service selection using multicriteria decision anal-
ing VM to minimize the energy consumption. Currently, ysis. Sci. World J. 2014, 1–10 (2014)
cloudSim is used as general framework to prove the suit- 2. Buyya, R., Yeo, C.S., Venugopal, S., Broberg, J., Brandic, I.: Cloud
computing and emerging IT platforms: vision, hype, and reality for
ability of models. We need to evaluate these models with
delivering computing as the 5th utility. Future Gener. Comput. Syst.
other type of application keeping uncertain workload 25(6), 599–616 (2009)
environment. 3. Zadeh, L.A.: The role of fuzzy logic in modeling, identification
• Extensive research will be conducted to virtual machine and control. Model. Identif. Control 15(3), 191–203 (1994)
4. Prasath, V., Nithya Bharathan, N., Lakshmi, N.: Fuzzy logic in
migration where all the models are applied in single
cloud computing. Int. J. Eng. Res. Technol. 2(3), 1–5 (2013)
cloud. That is, migrating the of VM occurs among the 5. Buyya, R., Broberg, J., Goscinski, A.M.: Cloud Computing Prin-
DCs of a specific cloud, we need to redesign the algo- ciple and Paradigms. Wiley, New Jersey (2011)
rithm that can applied in inter-cloud to solve the migration 6. Freed, D., Agrawal, R., John, S., Walker, J.J.: Cloud forensics chal-
lenges from a service model standpoint: Iaas, Paas and Saas,” In:
problems.
Proceedings of ACM International Conference on Management of
• Different models are developed to overcome the uncer- Computational and Collective Intelligence in Digital EcoSystem,
tainty and complexity between cloud service provider and pp. 148–155 (2015)
cloud service user regarding services suitability. Differ- 7. Sotomayor, B., Montero, R.S., Llorente, I.M., Foster, I.: Virtual
machine infrastructure management in private and hybrid clouds.
ent types of attributes are considered for different models.
IEEE Internet Comput. 13(5), 14–22 (2009)
Extensive research is required to overcome the mismatch 8. Subashini, S., Kavitha, V.: A survey on security issues in service
problems that may occur between cloud service provider delivery models of cloud computing. J. Netw. Comput. Appl. 34(1),
and user and expert technique are required to evaluate the 1–11 (2011)
9. Hayes, B.: Cloud computing. Commun. ACM 51(7), 9–11 (2008)
complete quality model.
10. Youseff, L., Butrico, M., Silva, D.: Towards a unified ontology
• Majority of research using fuzzy logic is conducted in of cloud computing. In: Proceedings of Grid Computing Environ-
the area of resource optimization, scheduling, and service ments Workshop, pp. 1–10 (2008)
dependability in cloud computing, But, they have ignored 11. Xue, J., Li, L., Zhao, S., Jiao, L.: A study of task scheduling based
on differential evolution algorithm in cloud computing. In: Pro-
the area of data storage and security in cloud computing
ceedings of IEEE conference on Computational Intelligence and
using fuzzy logic. Deep research is required to overcome Communication Networks, pp. 637–640 (2014)
the problems in data storage and security. 12. Alla, H., Alla, S.B., Ezzati, A., Mouhsen, A.: A novel architec-
ture with dynamic queues based on fuzzy logic and particle swarm
optimization algorithm for task scheduling in cloud computing. In:
El-Azouzi, R., Menasche, D., Sabir, E., De Pellegrini, F., Benjil-
7 Conclusion lali, M. (eds.) Advances in Ubiquitous Networking, vol. 397, pp.
205–217. Springer, Singapore (2016)
In this study, we focused on use of fuzzy logic in cloud 13. Pandey, S., Wu, L., Guru, S., Buyya, R.: A particle swarm
optimization-based heuristic for scheduling workflow applications
computing with different membership function and with dif- in cloud computing environments. In: Proceedings of IEEE Inter-
ferent type of methods. We describe the fuzzy logic based national Conference on Advanced information networking and
on their membership function with different type of meth- applications, pp. 400–407 (2010)

123
Cluster Comput

14. Xu, B., Wang, K., Wang, Y.: An improved artificial bee colony algo- 33. Albano, L., Anglano, C., Canonico, M., Guazzone, M.: Fuzzy—
rithm for cloud computing service composition. In: Proceedings of Q&E: achievement QoS guarantees and energy saving for cloud
IEEE International Conference on Heterogeneous Networking for application with fuzzy control. In: Proceedings of IEEE Interna-
Quality, Reliability, Security and Robustness, pp. 310–317 (2015) tional Conference on Cloud and Green computing, pp. 159166
15. Lee, B., Oh, K., Park, H., Kim, U., Youn, H.: Resource reallocation (2013)
of virtual machine in cloud computing with MCDM algorithm. In: 34. Toosi, A.N., Buyya, R.: A fuzzy logic-based controller for cost and
Proceedings of IEEE International Conference on Cyber-Enabled energy efficiency load balancing in geo- distributed data centers. In:
Distributed Computing and Knowledge Discovery, pp. 470–477 Proceedings of IEEE/ACM 8th International Conference on Utility
(2014) and Cloud Computing, pp. 186–194 (2015)
16. Qu, L., Wang, Y., Orgun, M.A.: Cloud service selection based 35. Ramezani, F., Naderpour, M., Lu, J.: A multi-objective optimiza-
on the aggregation of user feedback and quantitative performance tion model for virtual machine mapping in cloud data center. In:
assessment. In: Proceedings of IEEE International Conference on Proceedings of IEEE International Conference on Fuzzy Systems,
Services Computing, pp. 152–159 (2013) pp. 1259–1265 (2016)
17. Junior, R., Romlim, T.: A multi-criteria approach for assessing 36. Esposito, C., Ficco, M., Palmieri, F., Castiglione, A.: Smart cloud
cloud deployment options based on non-functional requirements. storage service selection based on fuzzy logic, theory, of evidence
In: Proceedings of the Annual ACM Symposium on Applied Com- and game theory. IEEE Trans. Comput. 65(8), 2348–2362 (2016)
puting, pp. 1383–1389 (2015) 37. Saripalli, P., Pingali, G.: MADMAC: multiple attribute decision for
18. Opricovic, S., Tzeng, G.H.: Compromise solution by MCDM meth- adoption of clouds. In: Proceedings of IEEE International Confer-
ods: a comparative analysis of VIKOR and TOPSIS. Eur. J. Oper. ence on Cloud Computing, pp. 316–323 (2011)
Res. 156(2), 445–455 (2004) 38. Qu, C., Buyya, R.: A cloud trust evaluation system using hierarchi-
19. Fan, C., An, C.: Employing the grey relational analysis to identify cal fuzzy inference system for service selection. In: Proceedings
and evaluate cloud computing risk. Int. J. Recent Res. Appl. Stud. of IEEE International Conference on Advanced Information Net-
15(1), 1–11 (2013) working and Application, pp. 850–857 (2014)
20. Zhao, J., Bose, B.: Evaluation of membership function for fuzzy 39. Hosseini, S.R., Adabi, S., Tavoli, R.: A near optimal approach in
logic controlled induction motor drive. Proc. IEEE Annu. Conf. choosing the appropriate physical machine for live virtual machine
Ind. Electron. Soc. 4, 229–234 (2002) in cloud computing. J. Adv. Comput. Eng. Technol. 1(3), 23–32
21. Nine, M., Azad, M., Abdullah, S., Rahman, R.M.: Fuzzy logic (2015)
based dynamic load balancing in virtualized data centers. In: Pro- 40. Lo, C., Tsai, C.F., Chao, K.M.: Service selection based on fuzzy
ceedings of IEEE International Conference on Fuzzy Systems TOPSIS method. In: Proceedings of IEEE International Conference
(2013) on Advanced Information Networking and Application Work-
22. Priya, V., Babu, C.: Moving average fuzzy resource scheduling shops, pp. 367–372 (2010)
for virtualized cloud data services. Comput. Stand. Interfaces 50, 41. Wu, H., Wang, Q., Wolter, K.: Methods of clouds-path selection for
251–257 (2017) offloading in mobile cloud computing systems. In: Proceedings of
23. Kong, X., Lin, C., Jiang, Y., Tan, W., Chu, X.: Efficient dynamic IEEE International Conference on Cloud Computing Technology
task scheduling in virtualized data center with fuzzy prediction. J. and Science, pp. 443–448 (2012)
Netw. Comput. Appl. 34, 1068–1077 (2011) 42. Su, C.H., Tzeng, G.H., Tseng, H.L.: Improving cloud computing
24. Kumar, V.V., Dinesh, K.: Job scheduling using fuzzy neural net- service in fuzzy environment-combining fuzzy DANP and Fuzzy
work algorithm in cloud environment. Bonfring Int. J. Man Mach. VIKOR with a new Hybrid FMCDM model. In: Proceedings of
Interface 2(1), 1–6 (2012) IEEE International Conference on Fuzzy Theory and Its Applica-
25. Grandhi, S., Wibowo, S.: Performance evaluation of cloud tion, pp. 30–35 (2012)
computing providing using fuzzy multiattribute group decision 43. Alabool, H.M., Mahmood, A.K.: Trust-based service selection in
making model. In: Proceedings of IEEE International Confer- public cloud computing using fuzzy modified VIKOR method.
ence on Fuzzy System and Knowledge Discovery, pp. 130–135 Aust. J. Basic Appl. Sci. 7(9), 211–220 (2013)
(2015) 44. Tajvidi, M., Ranjan, R., Kolodziej , J., Wang, L.: Fuzzy cloud ser-
26. Xu, J., Zhao, M., Fortes, J., Carpenter, R., Yousif, M.: On the use vice selection framework. In: Proceedings of IEEE International
of fuzzy modeling in virtual data center management. In: Proceed- Conference on Cloud Networking, pp. 443–448 (2014)
ings of IEEE International Conference on Autonomic Computing 45. Singla, C., Kaushal, S.: Cloud path selection using fuzzy analytic
(2007) hierarchy process for offloading in mobile cloud computing. In:
27. Baliyan, N., Kumar, S.: A hierarchical fuzzy system for quality Proceedings of IEEE International Conference on Recent Advances
assessment of semantic web application as a service. ACM SIG- in Engineering and Computational Sciences (2015)
SOFT Softw. Eng. Notes 41(1), 1–7 (2016) 46. Tarighi, M., Motamedi, S.A., Sharifian, S.: A new model of vir-
28. Barua, A., Mudunuri, L., Kosheleva, O.: Why trapezoidal and tri- tual machine migration in virtualized cluster server based on fuzzy
angular membership function work so well: towards a theoretical decision making. J. Telecommun. 1(1), 40–51 (2010)
explanation. J. Uncertain Syst. 8(3), 164–168 (2014) 47. Jaiganesh, M, Antony Kumar, A.V.: B3: fuzzy based data center
29. Sethi, S., Sahu, A., Jena, S.K.: Efficient load balancing in cloud load optimization in cloud computing. Math. Prob. Eng. (2013)
computing using fuzzy logic. IOSR J. Eng. 2(2), 65–71 (2012) 48. Xiaojun, W., Yun, W., Zhe, H., D. Juan, D.: The research on
30. Mehranzadeh, A., Hashemi, S.M.: A novel- scheduling algorithm resource scheduling based on fuzzy clustering in cloud computing.
for cloud computing based on fuzzy logic. Int. J. Appl. Inf. Syst. In: Proceedings of IEEE International Conference on Intelligent
5(7), 28–31 (2013) Computation Technology and Automation, pp. 1025–1028 (2015)
31. Minarolli, D., Freisleben, B.: Virtual machine resource allocation 49. Su, T., Wang, S., Vu, H., Ku, D., J. Huang, J.: An application
in cloud computing via multi-agent fuzzy control. In: Proceedings of fuzzy theory to the power monitoring system in cloud envi-
of IEEE International Conference on Cloud and Green Computing, ronments, In: Proceedings of IEEE International Symposium on
pp. 188–194 (2013) Computer, Consumer and Control, pp. 350–354 (2016)
32. Chen, Z., Zhu, Y., Di, Y., Feng, S.: A dynamic resource scheduling 50. Mukherjee, K., Sahoo, G.: Mathematical model of cloud comput-
method based on fuzzy control theory in cloud environment. J. ing framework using fuzzy bee colony optimization technique. In:
Control Sci. Eng. 2015, 10 (2015) Proceedings of IEEE International conference on Advances in com-

123
Cluster Comput

puter, control and Telecommunication Technologies, pp. 664–668 Kyong Hoon Kim received his
(2009) B.S., M.S., and Ph.D. degrees
51. Singhal, U., Jain, S.: A new fuzzy logic and GSO based load bal- in Computer Science and Engi-
ancing mechanism for public cloud. Int. J. Grid Distrib. Comput. neering from POSTECH, Korea,
7(5), 97–110 (2014) in 1998, 2000, 2005, respec-
52. Ramezani, F., Naderpour, M., J. Lu, J.: Handling uncertainty in tively. Since 2007, he has been
cloud resource management using fuzzy bayesian network. In: an associate professor at the
Proceedings of IEEE International Conference on Fuzzy Systems Department of Informatics,
(2015) Gyeongsang National Univer-
53. Monil, M.A.H., Rahman, R.M.: VM consolidation approach based sity, Jinju, South Korea. From
on heuristics, fuzzy logic, and migration control. J. Cloud Comput. 2005 to 2007, he was a post-
5(1), 1–8 (2016) doctoral research fellow at
54. Ooi, B.Y., Chan, H.Y., Cheah, Y.N.: Resource selection using fuzzy CLOUDS lab in the Department
logic for dynamic service placement and replication. Proceed- of Computer Science and Soft-
ings of IEEE Trends and Development in Converging Technology ware Engineering, the University
towards 2020, pp. 128–132 (2011) of Melbourne, Australia. His research interests include real-time sys-
55. Jamshidi, P., Ahmad, A., Pahl, C.: Autonomic resource provision- tems, cloud computing, and security.
ing for cloud-based software. In: Proceedings of ACM International
Symposium on Software Engineering for Adaptive and Self Man-
aging Systems, pp. 95–104 (2014)
Ki-Il Kim received the M.S.
56. Mon, M.A.H., Rahman, R.M.: Fuzzy logic based energy aware
and Ph.D. degrees in computer
vm consolidation. In: Proceedings of International Conference on
science from the Chungnam
Internet and Distributed Computing System, pp. 31–38 (2015)
National University, Daejeon,
57. AliPour, M.M., Derakhshi, M.R.F.: Two level fuzzy approach for
Korea, in 2002 and 2005, respec-
dynamic load balancing in the cloud computing. J. Electron. Syst.
tively. He is with Department of
6(1), 17–31 (2016)
Computer Science and Engineer-
58. Piegat, A.: Fuzzy Modeling and Control. Springer, Berlin (2013)
ing, Chungnam National Uni-
59. Ross, T.J.: Fuzzy Logic with Engineering Applications. Wiley, New
versity, Daejeon, Korea. Prior
York (2009)
to joining, he has been with
60. Ahmad, J., Siyal, M.Y., Najam, S., Najam, Z.: Fuzzy Logic Based
the Department of Informatics at
Power Efficient Real Time Multi-core Systems. Springer, Berlin
Gyeongsang National University
(2016)
since 2006. His research inter-
61. Mya, S., Thein, N.L.: A resource pool management model using
ests include routing for MANET,
fuuzzy logic decision making. Int. J. Comput. Appl. 29(10), 24–31
QoS in wireless network, multi-
(2011)
cast, and sensor networks.
62. Perumal, B., Murugaiyan, A.: A firefly colony and its fuzzy
approach for server consolidation and virtual machine placement
in cloud datacenters. Adv. Fuzzy Syst. 2016, 1–15 (2016). doi:10.
1155/2016/6734161

Bashir Hayat received BCS


(Hons) degree from University
of Peshawar, MS from Sha-
heed Zulfikar Ali Bhutto Insti-
tute of Science and Technol-
ogy (SZABIST) Islamabad and
enrolled as a Ph.D. researcher
at Gyeongsang National Univer-
sity, Real Time System (RTS)
Lab, Jinju, South Korea. His
research interests include Cloud
computing, Fog Computing, reso-
urce and energy optimization.

123

You might also like