Inﬂuential Reasonable Robust Virtual
Machine Placement for Eﬃcient
Utilization and Saving Energy
Bibi Ruqia1, Nadeem Javaid2(B
), Altaf Husain3, Najeeba Muhammad Hassan1,
Haﬁza Ghulam Hassan1, and Yumna Memon4
1Sardar Bhadur Khan Women University Quetta, Quetta 87300, Pakistan
2COMSATS Institute of Information Technology, Islamabad 44000, Pakistan
3Balochistan Universty of Information Technology and Management Sciences,
Quetta 87300, Pakistan
4International Dormitory, Wuhan University, Wuhan 430072, Hubei Province, China
Abstract. The integration of Cloud-Fog Platform (CFP) is built in
order to provide online services to the consumers in an eﬃcient way.
Dynamic changes of resources put load on servers. Due to which extra
energy demands and an improper usage of energy by consumers have an
eﬀect on the utility. Virtual Machine Placement (VMP) problem is con-
sidered to be solved with optimization technique as allocation of Virtual
Machines (VMs) to a single Physical Machines (PMs). The distribution
of energy with ineﬃcient utilization of resources causes of the energy deﬁ-
ciency in noticing daily updates of consumers in a month. In this paper,
game theory with coalition and non-coalition mechanism are applied for
purpose of balancing electricity load among consumers. Results show
that expectation of demanding electricity is kept low in order to mini-
mize improper way of utilization of energy. However, increment in saving
of energy will help consumers to sort out arising issue of an unbalanced
load on utility due to extra demand. The eﬃcient distribution of energy is
addressed in order to have proper utilization and management of energy.
Therefore, energy consumption is minimized due to eﬃcient utilization
Keywords: Cloud and fog platform ·Virtual machine placement
problem ·VM allocation ·Optimization technique ·Game theory ·
Cloud computing provides online services for users and utilization of virtual-
ization technology in computing environment to meet requirements of diﬀerent
applications to increase resource consumption . In , SG concept, renewable
Springer Nature Switzerland AG 2020
L. Barolli et al. (Eds.): IMIS 2019, AISC 994, pp. 549–561, 2020.
550 B. Ruqia et al.
energy sources adopt in the electric power industry. The demand side manage-
ment including SG for actualizing ability of SG. Author has described loads as a
shiftable appliances and cost of electricity minimizes on the basis of consumption
proﬁle. In , EMC designs to have control on power utilization of automatic
Researchers have discussed in  that objectives are needed to minimize con-
sumption without considering power utilization of physical edge which needs to
be addressed VMs assignments. Therefore, incoming tasks always run in VMs.
The time slot based resource allocation mechanism has mapped multiple VMs
in a single PM. Authors in  have discussed VMP which is an important prob-
lem to be resolved. Objective of VMP needs to have little consumption rate.
Recently, many authors have proposed some approaches to VM migration with
less involvement of the hypervisor between an original VM instance to be change
during execution . In this research work, management of VMs are discussed
in order to schedule daily consumption of consumers. The eﬃcient utilization
of energy matters to balance the load. The scheme is used to optimize utiliza-
tion of reasonable placement of VMs onto PM and data centers’ energy saving.
Nowadays, researchers are focusing in virtualization of server in order to smart
managements of VMs in PM. A robust VMP scheme is used to balance load in
VMs onto PM and the fully Utilization of resources. Multi objective VMP model
builds to balance load and minimize energy consumption and maximize load as
well as considering diﬀerent requirements of cloud providers.
Cloud data centers are faced with serious problem of increasing energy con-
sumption. VMP problem uses to balance the load on VMs and also introduce for
eﬃcient utilization of energy saving. The objectives are to minimize the energy
consumption and maximize load balance, resource utilization, and robustness.
In general, the allocation of resources to VMs and also little use of VMs in PMs.
Traﬃc ﬂows among VMs are very high due to data transfers in between multi-
ple PMs in distributing application, intensively. The control of VMs are given to
hypervisor, which is only a proper way of dealing with PMs and VMs. The hyper-
visor limits resources to each VM onto PM. VMs are managed in such a way that
the interference among each resource is not found under any condition and VMs
in a PM are considered independent. The eﬃcient conﬁguration helps assigning
the tasks that will not suﬀer from any delay. Furthermore, contributions are
presented below which are achieved up till now. The remainder of synopsis is
organized as follows: related studies are presented in Sect. 3. Section 4describes
the problem statement. System models along with the proposed solution are
demonstrated in Sects. 4and 5.
•Management of utilization and consumptions are considered with eﬃcient
•Consolidation of VM are focused to have minimum cost.
•Allocation of VMs onto PMs are noticed to balance loads.
•Placement of VM are considered to perform tasks.
Inﬂuential Reasonable Robust Virtual Machine Placement . . . 551
•Game theory are used as a mechanism for distribution and collision.
•In order to validate eﬀectiveness of our proposed scheme, simulations are
performed in MATLAB.
2 Problem Statement
In energy eﬃciency, VMs consolidation has an implication which leads to a VMP
problem, that mean allocation of lots of VMs to PM . A complex NP-Hard
problem is named as a VMP problem which seeks to obtain the best allocation
of VMs on PM [8,9]. Researchers have discussed diﬀerent approaches for VMP
objectives in the literature review. Such as, for energy eﬃciency, optimizing the
assignment of VMs on PM , VM consolidation to have increments in resource
utilization , and balance the load on many other PM to boost eﬃciency of the
whole system [12,13]. In cloud computing domain, in  applies an Ant Colony
Optimization (ACO) approach which brings the cloud servers into little amount
as a support for balancing the load. However, a high cost and consolidates VMs
on a single PM, are seen in this method. Further, in , authors have proposed
an Order Exchange and Migration Ant Colony System HOEMACS algorithm,
which incorporates in order to handle dynamic environment to allocate VMs for
reducing energy consumption. Memory utilization is not balanced in this scheme,
i.e., memory is bottleneck for heterogeneous environments. Therefore, we need
an eﬃcient system that can tackle this bottleneck.
3 Related Work
In , Home Energy Management System (HEMS) is proposed to ﬁnd the best
scheduling scheme for home appliance. Dynamic programming techniques have
been used with integration and coordination among appliances. Analyses vali-
date the eﬃcient scheduling of the appliances. However, a time based pricing
scheme is set for on and oﬀ peak hours to see high cost. Authors in have
proposed Centralized Approach To Mobile Adhoc Network (CATMAN) app-
roach which is integrated with Software Deﬁned Network (SDN). Researchers
have tried to optimize results while extracting out from new proposed routing
algorithm, however, the designed protocol runs when centralized control is miss-
ing which means logical control is needed to induce protocol work better than
In , algorithm has handled tenants eﬃciently and provided signiﬁcant
gains. However, results of Layered Progressive resource allocation based on Mul-
tiple Knapsack Problem (LPMKP) has shown few abnormal cases which demon-
strates fewer diﬀerences in Quality of Service (QoS) as compared to other algo-
rithms. Authors in  have focused on a coalition mechanism in Geo-distributed
Mobile Cloud Computing (GMCC) network which raises cooperation between
remote and local Service Providers (SPs). Further, an improved coalition app-
roach has also been introduced and applied that helps to balance other SPs with
low resource utilization and needs more resources to run applications. However,
552 B. Ruqia et al.
resources are in suﬃcient to provide services. Cost of data transmission and
migration are increased in case of increment in cooperation also.
Finite computing resources are to be optimized. Therefore, researchers have
focused on Multi Objectives Scheduling ACO (MOSACO) algorithm in . This
algorithm provides the highest optimality according to consideration of task
completion time, service providers, cost, and QoS. In VM migration, decision
making is essential because of large amount of VMs in PM which is considered
as a complexity in planing of VMs migration. Hybrid resource pool model is pro-
posedin to bring reduction in migration. Authors have improved eﬃciency
through maximum utilization of hirepool. The results got improved by reducing
In , authors have implemented a distributed Network Virtualization
Hypervisor (NVH): DPVisor is virtualized and dispatched message to Vir-
tual SDN (VSDN), whenever DPVisor receives a message from data plane.
Researchers have discussed optimization of distributed operations. However, all
instances of correct updates realize VSDN slicing essentially. A ﬂow table of
consumption for switches to process VSDN is proposed and protocol stack is
developed as an interface.
Authors  have proposed a Service-oriented VMP (SVMP) strategy which
is a suitable intelligent computing platform of Internet of Thing (IoT) back end
and also divided the roles of VM into Web role, worker role, and storage accord-
ing to function types. The genetic algorithm is used to conduct the optimal
conﬁguration for diﬀerent types of VMs under the situation of limited resources
in order to achieve minimum communication overhead and total power consump-
tion. The results have proved that the proposed placement strategy has achieved
higher eﬃciency in data centre that minimizes an unnecessary resource waste.
However, mutation rate increases because of increase in environmental scale of
An algorithm schedules tasks and allocates resources by Heuristic Cloudlet
Allocation (HCA) algorithm in . An algorithm has acquired scheduling of task
for balance load and less makespan beside limitation in resource capacity. Het-
erogeneous resources are being utilized completely as a result an algorithm has
the best eﬀect on balancing for arranging huge cloudlets with ﬁxing limitation.
However, an algorithm has dealt only with dependent cloudlets to minimize com-
munication cost. Researchers have discussed increment of energy consumption
in cloud environment while executing task. Researchers aim to apply proposed
approach for dynamic application and also select the best VM for execution and
oﬄoading tasks. Authors have discussed in ﬁrst layer for realizing the fastest
conﬁrmation of data and policy packages for providing a certain integrity.
A new approach Consensus Achievement Algorithm (CAA) in  represents
measurement of VM storage location and assure to secure storage in cloud. Fur-
ther, researchers have thought about block chain that has multiple settings to
use functions’ dependent work on searching an eligible nonce. The bitcoin sys-
tem generates only tamper-resistant meta data. This systems’ amount of data
is small as compared to size of VM measurements polices. However, total com-
Inﬂuential Reasonable Robust Virtual Machine Placement . . . 553
puting power is ﬁxed, time of building block, and maintaining block by varying
diﬃculties for practical. Authors have introduced technique of integrating share
services for resources and have made general purpose computing on Graphics
Processing Unit (GPU) to demonstrate beneﬁts in production environments.
The fogs have PMs which have hypervisors. There are three PMs which are exist-
ing inside the fog. The hypervisor is controlling many Virtual Machine Monitors
(VMMs). The VMMs are responsible for each VMs. The groups of ﬁve VMs are
made in PMs by VMMs. With each group of VMs in PMs, the limited resources
are available. The resources are divided into diﬀerent tasks. The limited archives
are ﬁxed to each PMs inside the fog. The archives help to restore and retrieve
the data as information for the cloud. Now, each PMs have groups of VMs with
limited resources and also having a storing facility also. The fogs work eﬃciently
due to utilizing resources in an organized way to fulﬁll the request of any user.
The users are mentioned by their apartments in Fig. 1. Each apartment has a
proﬁle in a fog.
Fig. 1. The system model of cloud, fog, and utility having a scenario description inside
The demand for energy is done by requesting the fog. The utility generates
energy for users however energy will be supplied by the energy demand. The
providing of energy is depending on the surplus of energy. The users must be
eﬃcient to save energy in order to be the supplier as well. Each apartment has
the battery to save energy. It depends on users how much they can save energy.
The details of all the batteries attached to the apartments are on the fog. The
utilization of the energy put an eﬀect on the energy storage. There is the surplus
554 B. Ruqia et al.
of energy because the energy is generated for apartments more than enough,
however, the energy will be supplied according to their need. Some apartments
have stored the energy even that limited energy which will help them earning
money being supplier. Some apartments have not utilized the energy well, now
that apparently suﬀering from not having any. Here, the storage will help those
apartments which are suﬀering from energy deﬁciency.
The apartments which have the surplus of energy will serve those apartments
who do not have energy. These apartments are found being energy generator as
well through diﬀerent renewable energy resources. Now, the need is collaborat-
ing with each other because after utilizing the limited energy of themselves, the
apartments are compelled to contact others. The restriction on each apartment
needs as providing the services through agreement. The MGs are existing near
the apartments which provide energy to the apartments and also deal with such a
customer who is suﬀering from energy deﬁciency. The proﬁle shows those apart-
ments which have energy surplus because the MGs are having limited energy to
distribute the limited amount of energy from diﬀerent apartments. The solution
is to cooperate with each other to release from such kind of daily problems. In
case of not collaborating the energy demand increases which burden the MGs to
request fog further energy demand. The cooperative and non-cooperative users
are enlisted in the proﬁle of MG separately. These apartments will be checked out
through their proﬁles. The MGs keep records of each apartment and forwarding
all details to fog on each progress.
Figure 2shows the electricity load with respect to time. The proﬁle of a month
shows each day’s usage of electricity load on utility. The users are categorized
in apartments based on their apartment requirement. There are ten apartments
and use of electricity is shown by their proﬁles. The range of electricity load
is shown from 0 to 6000. The electricity usage of apartment 4, 7, and 10 are
lowest form of all other apartments. The electricity load is shown cooperation
and non-cooperation of users in Fig. 3. Usage is captured for 24 h in a day. The
generation of electricity with respect to users who are balancing load of utility
with their cooperation is shown better than not cooperating with each other.
The extra demand for electricity with wastage of energy resources shows in
Fig. 4. Fluctuation of lines shows wastage and extra demand. The ﬁrst three
hours show the best result because load is balanced and no energy wastage and
extra demand has appeared. The extra demand is the highest on day 12 and 20,
which is even more than wastage. Whenever the lines seem high in range, the
wastage and extra demand will be high too. Also, extra demand shows higher
on some days without wastage of energy. The range seems high as much wastage
of energy is high. In Fig. 5, the electricity load shows per day. Whenever there
is extra generation of energy there is surplus of energy and energy isn’t saved
however, extra demand is also shows high. The electricity load increases six times
higher in six days.
Inﬂuential Reasonable Robust Virtual Machine Placement . . . 555
1 2 3 4 5 6 7 8 9 101112131415161718192021222324252627282930
Fig. 2. The proﬁle of ten apartments’ daily loads for a month
1 2 3 4 5 6 7 8 9 101112131415161718192021222324252627282930
Electricity Load (kW)
Micro grid Extra Load Demand Surplus Energy
Fig. 3. The extra load demand from MG and surplus of MG’s energy show per day
electricity load on the utility
As a result, there is surplus of energy that is saved and other days show that
there are extra generation on demand and energy is not saved. The energy is
wasted as not being useful utilization. If action performs on utilization, extra
generation will not be need. Though ﬂuctuations clearly shows surplus of energy
which is not being saved on the daily basis. The extra demand is high that is
why extra load generation requests are high on utility.
Figure 6shows the proﬁle of a month where the load is exhibited with
respected to time. The expected demand and utilization are huge in range. The
extra generation and extra demand seem low among all lines. In some days, it
is showing that the surplus energy is saved, however, mostly it is utilized that
is why the utilization rate is high. The usage of energy shows in Fig. 7that
how energy is being used by coalition and without coalition during all day. The
556 B. Ruqia et al.
Electricity Load (kW)
Extra demand Energy wasted
Fig. 4. Extra demand and energy wastage show per day electricity load on the utility
Electricity Load (kW)
Expected Demand Utilized Surplus Extra Demand Save Extra Generation
Fig. 5. Electricity load shows while expected demand, utilization, surplus, extra
demand, save, and extra generation are noticed with respect to time
generation is denoted by G. The energy wasted is indicated by EL. The deﬁcient
energy is represented by ED. The extra energy of MG generation without coali-
tion by LD. The extra energy MG generation with coalition signify by EMG.
The generation of energy grows highly, however, still there is the availability of
energy eﬃciency due to energy wasted. The rate of wasted energy is higher than
the other two bars which show the deﬁciency of energy and that is due to not
cooperating with each other. The rate of deﬁciency is equal to the LD bar. The
EMG is showing very low rate than all others.
Inﬂuential Reasonable Robust Virtual Machine Placement . . . 557
Electricity Load (kW)
Generation WOC WC
Fig. 6. Energy generation with respect to cooperation and without cooperation scheme
Fig. 7. Extra load demand and surplus energy of MG shows per day electricity load
on the utility
Figure 8discloses a complete load of electricity on daily requests of users.
The loads are displayed for certain range on each day. The range of the loads is
existing in between 0 and 7000. Only day 25 touches the origin of zero, however,
all the others’ day points are existing upward. In Fig. 9, bars are exposed loads
and that load’s range reached near 12,000. Range is still growing which is counted
558 B. Ruqia et al.
Electricity Load (kW)
Extra Energy MG G WOC
Extra Energy MG G WC
Fig. 8. Load proﬁle shows each day performance
1 2 3 4 5 6 7 8 9 101112131415161718192021222324252627282930
Electricity Load (kW)
Fig. 9. Cooperative and non-cooperative game theory’s eﬀect on energy usage to know
the energy deﬁciency due to wastage
as per day. The small bar in plot shows the lowest one. In Fig. 10, bars are existing
above origin of zero which shows balancing of load properly. However, very little
amount of bars are shown in opposite side of origin zero which means loads still
need to be balanced completely.
Inﬂuential Reasonable Robust Virtual Machine Placement . . . 559
Fig. 10. Load is highly balanced after many iterations with the lowest and highest
In this paper, the CFP provides services to the building according to the need
of consumers’ demand by scheduling their daily updates. The building is hav-
ing multiple apartments attached with MG to have electricity from utility. The
energy generation is the largest among all opponents. The results show the
energy is managed in a better way due to coalition mechanism, however, the
non-coalition mechanism among the consumers has shown energy deﬁciency.
The performance of coalition mechanism is better than the non-collision. The
energy deﬁciency demonstrates equal to non-coalition mechanism and the energy
wasted is larger energy deﬁcient. however, the increment in performance with
method coalition removes the energy wastage issue with extra demand while
having the surplus energy.
1. Sahoo, P.K., Dehury, C.K., Veeravalli, B.: LVRM: on the design of eﬃcient link
based virtual resource management algorithm for cloud platforms. IEEE Trans.
Parallel Distrib. Syst. 1, 1 (2018)
2. Samuel, O., Javaid, S., Javaid, N., Ahmed, S., Afzal, M., Ishmanov, F.: An eﬃcient
power scheduling in smart homes using Jaya based optimization with time-of-use
and critical peak pricing schemes. Energies 11(11), 3155 (2018)
3. Khalid, R., Javaid, N., Rahim, M.H., Aslam, S., Sher, A.: Fuzzy energy manage-
ment controller and scheduler for smart homes. Sustain. Comput. Inf. Syst. 21,
4. Wang, S., Xu, J., Zhang, N., Liu, Y.: A survey on service migration in mobile edge
computing. IEEE Access 6, 23511–23528 (2018)
5. Fatima, A., Javaid, N., Sultana, T., Hussain, W., Bilal, M., Shabbir, S., Ilahi, M.:
Virtual machine placement via bin packing in cloud data centers. Electronics 7(12),
560 B. Ruqia et al.
6. Wen, Y., Li, Z., Jin, S., Lin, C., Liu, Z.: Energy-eﬃcient virtual resource dynamic
integration method in cloud computing. IEEE Access 5, 12214–12223 (2017)
7. Bianchini, R., Rajamony, R.: Power and energy management for server systems.
Computer 37(11), 68–76 (2004)
8. Vogels, W.: Beyond server consolidation. ACM Queue 6(1), 20–26 (2008)
9. Greenberg, A., Hamilton, J., Maltz, D.A., Patel, P.: The cost of a cloud: research
problems in data center networks. ACM SIGCOMM Comput. Commun. Rev.
39(1), 68–73 (2008)
10. Feller, E., Rilling, L., Morin, C.: Energy-aware ant colony based workload place-
ment in clouds. In: Proceedings of the 2011 IEEE/ACM 12th International Con-
ference on Grid Computing, pp. 26-33. IEEE Computer Society, Sept 2011
11. Xiao, Z., Chen, Q., Luo, H.: Automatic scaling of internet applications for cloud
computing services. IEEE Trans. Comput. 63(5), 1111–1123 (2014)
12. Sahu, Y., Pateriya, R.K., Gupta, R.K.: Cloud server optimization with load bal-
ancing and green computing techniques using dynamic compare and balance algo-
rithm. In: 2013 5th International Conference and Computational Intelligence and
Communication Networks, pp. 527-531. IEEE, Sept 2013
13. Amokrane, A., Zhani, M.F., Langar, R., Boutaba, R., Pujolle, G.: Greenhead: vir-
tual data center embedding across distributed infrastructures. IEEE Trans. Cloud
Comput. 1(1), 36–49 (2013)
14. Ferdaus, M.H., Murshed, M., Calheiros, R.N., Buyya, R.: Virtual machine consol-
idation in cloud data centers using ACO metaheuristic. In: European Conference
on Parallel Processing, pp. 306-317. Springer, Cham, Aug 2014
15. Liu, X.F., Zhan, Z.H., Deng, J.D., Li, Y., Gu, T., Zhang, J.: An energy eﬃcient
ant colony system for virtual machine placement in cloud computing. IEEE Trans.
Evol. Comput. 22(1), 113–128 (2018)
16. Khalid, A., Javaid, N., Guizani, M., Alhussein, M., Aurangzeb, K., Ilahi, M.:
Towards dynamic coordination among home appliances using multi-objective
energy optimization for demand side management in smart buildings. IEEE Access
6, 19509–19529 (2018)
17. Siraj, M.N., Ahmed, Z., Hanif, M.K., Chaudary, M.H., Khan, S.A., Javaid, N.: A
hybrid routing protocol for wireless distributed networks. IEEE Access 6, 67244–
18. Yu, R., Ding, J., Maharjan, S., Gjessing, S., Zhang, Y., Tsang, D.H.: Decentralized
and optimal resource cooperation in geo-distributed mobile cloud computing. IEEE
Trans. Emerg. Top. Comput. 6(1), 72–84 (2018)
19. Zuo, L., Shu, L., Dong, S., Chen, Y., Yan, L.: A multiobjective hybrid cloud
resource scheduling method based on deadline and cost constraints. IEEE Access
5, 22067–22080 (2017)
20. Song, Q., Wang, X., Qiu, T., Ning, Z.: An interference coordination-based dis-
tributed resource allocation scheme in heterogeneous cellular networks. IEEE
Access 5, 2152–2162 (2017)
21. Xiong, R., Li, X., Shi, J., Wu, Z., Jin, J.: HirePool: optimizing resource reuse based
on a hybrid resource pool in the cloud. IEEE Access (2018)
22. Huang, H., Niu, B., Tang, S., Li, S., Zhao, S., Han, K., Zhu, Z.: Realizing Highly-
Available, Scalable, and Protocol-Independent vSDN Slicing with a Distributed
Network Hypervisor System (2018)
23. Zhang, L., Li, J.: Enabling robust and privacy-preserving resource allocation in fog
computing. IEEE Access 6, 50384–50393 (2018)
Inﬂuential Reasonable Robust Virtual Machine Placement . . . 561
24. Mapetu, J.P.B., Chen, Z., Kong, L.: Heuristic cloudlet allocation approach based
on optimal completion time and earliest ﬁnish time. IEEE Access 6, 61714–61727
25. Zhao, B., Fan, P., Ni, M.: Mchain: a blockchain-based VM measurements secure
storage approach in IaaS cloud with enhanced integrity and controllability. IEEE
Access 6, 43758–43769 (2018)