INTERNATIONAL JOURNAL OF SCIENTIFIC & TECHNOLOGY RESEARCH VOLUME 8, ISSUE 11, NOVEMBER 2019 ISSN 2277-8616

An Effective Owl Search Based Optimized

Resource Allocation Framework For Network
Slicing In An LTE Network
M. Leeban Moses, B. Kaarthick

Abstract: Network slicing enables service providers to set up multiple independent virtual networks to support a wide range of services and applications
on a single physical network. Network operators can also provide an expense-effective solution to satisfy different technical requirements of distributed
software and services with network slicing. Many of the wireless networks slicing strategies were computationally expensive due to the time it takes to
calculate the required resources for each individual slice. Hence, there seems to be no guarantee that a slice will equally share the resource allocation
between users. Every slice of the network needs two resource types, bandwidth as well as energy to process. As a result, resource distribution between
these two elements must be balanced, and conceptions of equality and efficiency become much more complicated. Therefore, in this paper, we intend to
propose an efficient network slicing resource allocation system on LTE network. Software Defined Networking (SDN) and Virtual Network (VN) can act
as network slicing building blocks by enabling network programmability. First, we best describe the LTE network model of the distribution system and
then the solution proposed. In addition, we use the MATLAB to analyze different scenarios to test the proposed models in this paper, and the simulation
results show that the proposed algorithm can substantially satisfy end-user resource allocation approach.

Index Terms: Network slicing, Software Defined Networking (SDN), Network Function Virtualization (NFV), LTE network, Resource management
——————————  ——————————

1. INTRODUCTION Considering the remarkable success of introducing

CELLULAR document networks have innovated virtualization technologies to tackle specific processing and
telecommunications over the past two decades and thus are computing problems in the IT sector, the research community
responsible for linking more individuals and phones than it has is investigating the cloudification of Internet-based services
ever been [1]. The exponential developments in mobile with the aim of creating network services that are autonomous,
network technology eventually tackled many networking scalable and secure [10]. Recent progress in networking
challenges; enhanced customer numbers and large amounts technology is reinforcing the case for cellular network
of data traffic have been handled [2]. With the number of users virtualization by Software Defined Networks (SDN), Network
drastically increasing the multimedia content consumption Function Virtualization (NFV), and Cloud-based Radio Access
specifically video and data traffic has been increased by Networks (CRAN). In addition to providing scalable network
something like of about 65% [3]. Because of the proliferation architecture, we agree that these technologies of virtualization
of cellular technology in the developing world, demand for can significantly reduce capital and operating costs [11].
bandwidth-intensive services such as video streaming, and the Consequently the Advanced virtualized networks have to get a
rise in machine-to-machine traffic, this trend is expected to new management mechanism to ensure resource allocation
continue [4]. Reports also estimate that the total number of reliability and assured isolation of resources [12]. To achieve
Internet-connected devices will hit 50 billion by 2020 out of just these goals, a new resource management system is needed
12.5 billion in 2010 [5]. The unprecedented increase in the that takes into account both low and high-level resource
percentage of mobile devices leads to massive greenhouse allocation management models [13]. In this paper we suggest
emissions due to heavy use of electrical energy from which a comprehensive approach to improve the way networks are
network infrastructure absorbs up to 90 percent [6]. The sliced and resource distribution is optimized by using the
mobile networks currently exist in the real world require power of software defined networking (SDN) and network
substantial changes in architecture to efficiently accommodate virtualization (NV) technologies. The key factors contributing to
the anticipated rise in user base and resulting in high demands the rapid adoption of network virtualization are: cost-effective
for bandwidth while at the same time reducing levels of energy network capacity sharing and high network usage. To achieve
usage [7]. Existing networks are also strongly hardware- unique and powerful incentives from network virtualization,
dependent, making the upgrade process for new wireless and together with the development of efficient network
access technologies both expensive and sluggish [8]. Experts architectures, we focus on an effective resource management
and operators have been looking at network virtualization to method using the metaheuristic approach in a virtual network
tackle these issues in cellular networks [9]. in this paper. The main role of the low-level model would be to
______________________ assign PRB-based resources in the number of units, thus
 Bannari Amman Institute of Technology, Sathyamangalam, ensuring high accuracy. The high-level design, on the other
India hand, should be responsible for ensuring independence
 Email: [email protected] among the resources allocated. Software Defined Network
 Coimbatore Institute 0f Engineering and Technology, (SDN) and Network Function Virtualization (NFV) open up new
Coimbatore, India opportunities to promote such versatile resource allocation,
adaptive network configuration and cost-effective network
operations [14]. The rest of the paper is structured as follows.
Section II offers some additional information on network slicing
used to update some of the existing research on digital
2581
IJSTR©2019
www.ijstr.org
 INTERNATIONAL JOURNAL OF SCIENTIFIC & TECHNOLOGY RESEARCH VOLUME 8, ISSUE 11, NOVEMBER 2019 ISSN 2277-8616

resource allocation. Section III explains the design of the inter-slice isolation and intra-slice isolation. Also developed
framework and the strategy proposed. Complete simulation was a generalized Kelly mechanism (GKM) based on which
results are given in Section IV and We conclude this paper in the resource allocation problem was discussed at the top
Section V. level. Another advantages of using such a resource bidding
and allocation system was that the seller (InP) did not need to
2 RELATED WORK: A BRIEF REVIEW know the true bidders (MVNOs) valuation. The optimal
Detailed The network slicing principle makes the resource allocation was extracted from each MVNO to its
implementation of LTE network services versatile and mobile users using KKT conditions to solve the lower resource
dynamic. Some of the works have been listed here in this allocation problem (i.e. between MVNOs and their mobile
section among the numerous past research works on resource users). Then there is the bandwidth resource to allocate the
management in network slicing. In a heterogeneous cloud MVNOs to the users. A. Fendt et.al [19] introduced a
infrastructure, Halabian et.al [15] introduced a resource mathematical model formalized as a structured Mixed Integer
allocation framework for virtualized 5 G networks. For each of Linear System to solve the offline Network Slice Embedding
its virtual network features, each network slice has a resource Problem. A latency-sensitive purpose feature ensures
demand vector in its design. First, they considered a maximum use of the network and low latency in network slice
cooperative slice model and formulated the resource allocation interactionL. Tang et.al [20] suggested a slice-based
as a problem of convex optimization, optimizing the overall scheduling of digital assets with NOMA software to increase
utility function of the system. They also proposed a the system's quality of service (QoS). They formulate the
decentralized solution to the problem of resource allocation by allocation of power granularity and subcarrier allocation
establishing a resource auction between the slices and the strategies into a question of the Constrained Markov Decision
data centers. By using an example, we demonstrate how the Process (CMDP), aimed at optimizing the overall user
egotism of non-collaborative slices affects the system's frequency. They developed an adaptive resource allocation
fairness quality. They also developed a new resource algorithm based on Approximate Dynamic Programming (ADP)
allocation question based on the notion of dominant resource to solve the problem in order to further escape the
equality for a process of non-collaborative slices and dimensionality curse and the expectation estimation in the
suggested a fully distributed scheme to solve the problem. optimal value function. G. Wang et.al [21] examined network
Architecture for the distribution of centralized radio assets has slice dimensioning with resource pricing policy. They initially
been suggested by Y. Sun et.al [16] in the paper, , in which the design a network slice dimensioning optimization model in
GRRM appeared to be accountable for allocating sub- which the problem of the Slice Customer (SCP) maximizes the
channels in slices to local radio resource managers (LRRMs), benefit of the SC and the problem of the Slice Provider (SPP)
who can then distribute the resources allocated to their UE. A maximizes net social welfare (resource efficiency). They
hierarchical resource allocation problem was developed under actually find that maximizing net social welfare and benefit
the architecture, and the problem was further modeled as a from SP are two clear priorities when resources are limited;
Stakelberg game with the GRRM as the leader and the LRRM and so a tradeoff occurred. Based on this result, they
as followers, taking into account the relationship between the proposed a distributed algorithm of low complexity in order to
GRRM and the LRRM. Due to the NP-hardness of the accomplish near-optimal net social welfare with a benefit
problems of the followers, a method was first proposed to guarantee for SP / SCs. J. Chen et.al [22] introduced a
achieve the Stackelberg equilibrium (SE) based on exhaustive dynamic resource management strategy called the Enterprise
search. Nonetheless, once the network size became relatively Visor engine which handles network resource allocation
large, it was impossible for game players to achieve SE within between slices. The EnterpriseVisor engine first slices the
minimal decision-making time. GRRM and LRRMs were seen network of an enterprise into virtual subnet and then controls
as restricted rational players facing this challenge, and low- the use of each subnet's asset. The Linear Programming
complexity algorithms were developed to help them achieve approach is used in the engine by dynamic resource
optimal local solutions leading to a poor SE version. K. Teague scheduling of the subnet for each operation. The engine then
et.al [17] explored the challenge of selecting base stations provides the schedule to data plane users as the FlowVisor
(BSs) to install a virtual network that fulfills a service provider's layer's resource allocation rule. A four-subnet OpenFlow
specific requirements and adaptive resource sharing between testbed was designed to demonstrate the feasibility of the
the demand points of the service provider. For modeling the proposed EnterpriseVisor engine.
problem of joint BS selection and adaptive slicing, a two-stage
stochastic optimization system was implemented. To evaluate
an approximation for the two-stage stochastic optimization
3 PROPOSED METHODOLOGY
model, two methods were presented. The first method uses a 3.1 LTE Network
sampling technique applied to the stochastic model's Number Here, we address the problem of network slice
determinist equivalent system. The second method uses a BS resource allocation in an LTE network in this paper. Our
choice and adaptive slicing genetic algorithm through a single- suggested method allows the user to push the device to any
stage linear optimization problem. A number of scenarios have functional traffic-fairness or computing-fairness trade-off. LTE
been developed for testing using a lognormal model designed network has become a current generation of cellular networks,
to emulate demand from cellular networks in the real world. Y. split into two parts: the LTE part deals with radio access
Tun et.al [18] discussed the issue of two-level allocation in technologies, meanwhile the the evolved packet core (EPC)
network slicing, thus allowing for efficient use of resources; deals with core network technology [23]. User device (UE)
2582
IJSTR©2019
www.ijstr.org
 INTERNATIONAL JOURNAL OF SCIENTIFIC & TECHNOLOGY RESEARCH VOLUME 8, ISSUE 11, NOVEMBER 2019 ISSN 2277-8616

connects via radio interface to an eNodeB. Perhaps the that slices of the network transmit the desired level of QoS to
eNodeB executes tasks for controlling radio resources such as an admission control body. Depending on the slice's
allocating radio resources and handling inter-cell interference. preference, the admission control mechanism chooses to
The eNodeB links via a mobile backhaul network to a serving serve the slice. Finally, the virtual network allocates the
gateway (SGW). The SGW supports a significant number of resources to the UEs / user of the accepted slices according to
eNodeBs and for the inter-eNodeB handover functions as the inter and intra slice priorities. As per the admission control
regional mobility anchor point. The SGW is linked to a Packet decision, the resource allocation function is carried out with
Data Network (PDN) gateway (PGW), which allocates UE the goal of optimizing the users ' quality of experience (QoE)
(user) IP addresses, enforces regulation, and filters packets within each slice, taking into account the inter-slice priority.
and charges. It is indeed a destination point to the internal
network for the packet data interface.
Resource information

UE Virtual Network slicing

Admission control mechanism

Slice allocate
BTS eNodeB
2G Node eB
3G

Resource required

LTE
BSC RNC Slicer

Mobile Backhaul

Slice allocate
SDN

to UE
SGSN HSS
MME

SGW

PCRF

PGW

Physical layer
Internet Operator IP
Service(IMS)

Fig.1. Basic Structure of an LTE network Fig.2. Proposed Resource Allocation Scheme

The main control body is the Mobility Management Entity The QoE is calculated in this paper by considering the users'
(MME), which is responsible for upholding mobility effective throughput, standardized according to their total
management conditions for UEs and setting up client traffic required data rate. To this end, the resources allocated to a
bearers. A central database where user profiles are stored is low-priority slice could be reduced, if required, to the minimum
the home subscriber server (HSS). It is responsible for amount capable of meeting the basic QoS criteria to allow
authenticating and approving the UE. The role of policy and higher priority for new slice(s). Our concept adjusts the sum of
charging rules (PCRF) provides the PGW with QoS and network resources assigned to network slices dynamically.
charging rules. User data packets are transmitted between The suggested access control mechanism progressively sets
eNodeB and PGW through GTP (GPRS tunneling protocol) the allocated assets to allow slices based on the current traffic
tunnels. Figure 1 demonstrates the overall LTE mobile network load. Consideration was provided to the inter-slice and intra-
structure. Whereas the LTE rolled off to accommodate with slice priority order to model the resource allocation task's QoE
mobile device development, it has some many challenges and maximization problem. In view of priority orders for QoE work,
opportunities [24]. The LTE is difficult to manage in radio the satisfactory rate of UEs can be improved network usage.
access and the mobile backhaul network, the installation of The design consists of four main elements: the system slice
many base stations and the control plane delivery, resulting in layer, the virtual layer of the network, the physical resources
inconsistent utilization of radio assets. With no clear and the admission control manager. The service slices provide
separation between control and data planes, the network is various services that demanded resources to be served. We
difficult to manage and monitor in the cellular core network. show in the virtual network with the collection of slices
The use of proprietary hardware by network organizations S  1, 2 , 3 ,......., S  . That slice has a set of UEs, a set like this is
leads to high implementation and operating costs and denoted by U s  1, 2 , 3 ,......., U s  . In terms of QoS constraints,
inefficient distribution and use of resources; this slows down that slice executes an application for admission control. We
the introduction of new innovations on the market and has an model a objective in this paper with R sm in and R sm ax , which
impact on operators ' profits [25]. denotes the minimum and maximum data rates related with
the slice. Every slice is prioritized  s , where such priorities are
specified with the restriction that   s  1 . Likewise, each
s S
3.2 Proposed Network Slicing Resource allocation client u belonging to the slice S , i.e., where  us 1
  #U ,
mechanism the virtual network layer provides an abstraction of the s s

The underlying mechanism for regulating acceptance takes full physical network resources, U s is defined by a priority  u . s

advantage of tier interests. Our concept is based on the idea The virtual network cuts the network infrastructure to
accommodate different slices according to the admission
2583
IJSTR©2019
www.ijstr.org
 INTERNATIONAL JOURNAL OF SCIENTIFIC & TECHNOLOGY RESEARCH VOLUME 8, ISSUE 11, NOVEMBER 2019 ISSN 2277-8616

control / control manager's decisions. The digital network this constraint often explicitly applies to the orthogonality of
receives requests from various slices in terms of the UEs to be delegated services. Framework (7b) indicates that the data
served for each slice and carries out the resulting allocation of rate obtained by UE us is constrained by the corresponding
physical resources according to the inter-and intra-slice slice s requirements. It should be noted that in Equation (6),
priorities while taking into account the UEs ' QoE, with this the QoE is a number lower than or equal to 1; as a result, the
aim, we can define higher a slice's priority, the lower the value of πs. This also
ru applies to the users, i.e. the higher a user's priority, the lower
 (
s
qu
s m ax
) the μus value. The resource allocation process is achieved by
Rs
optimizing the data rate using Owl search optimization
(1) algorithm as well as the UEs channel conditions.
As UE's QoE in the slice s; is UE's u slicing s data rate ru . s
The total slice client QoE can be estimated as: 3.3 Owl Search Algorithm
u The owl search algorithm launches the optimization process
qs   (q u )
s
s

by an initial set of random data rate solutions that reflects the

u s U s
forest owl's initial position (dimensional search space).
(2) Therefore to assign the initial position of each owl in the forest,
We can clearly describe that, a uniform distribution is used.

Q   (q s ) s
O i  O L  U ( 0 ,1 ) * ( O u  O L )
s S (8)
(3) Where ith owl O i in the jth dimension and U(0,1) are lower and
As, the total QoE that all UEs of all slices encountered, on a upper limits of O L and O u , respectively, is a uniformly
scheduling basis, the virtual network provides services. In a distributed random number in the range[0,1 ]. The fitness of
t
standard scheduling frame t, we define with, t
q us , q st and Q the position of each owl in a forest is evaluated and stored
the QoE. The time-average QoE values can therefore also be using an objective function in equation 8. It is presumed in the
defined as follows: present work that the fitness value of the location of each owl
1
directly relates the frequency information received through the
t
E [ q us ]  q us ears. Therefore, the strongest owl achieves maximum intensity
T
(for problems with maximization) as it is closer to the vole. The
(4) ith owls structured strength data is used to modify the location
1 t and can be measured as:
E [q s ]  qs
T fi  w
Ii 
(5) b  w
1 t (9)
E [Q ]  Q
T where b  f
(6) max k
k  1 ,.... n

Where T was the total number of scheduling frames w  f

considered, the physical resources correspond to the virtual
min k
k  1 ,.... n

network's radio infrastructure. For convenience's purpose, we Every single owl and prey's distance information is measured
refer to one vector-cell's backhaul network. B MHz represents with the following equation:
the total available bandwidth. The set M= { 1, 2,..., M} specifies R i  || O i , V || 2
the sub-channels available in which the standard sub-channel
(10)
m bandwidth is bm= B / M. The maximum transmission power
where V is the prey position obtained by the fittest owl, it has
PTOT is distributed uniformly to each sub-channel, i.e. pm= P /
also been stated that in the forest there is only one vole
M. The essential problem in concern during the resource
(global optimum). Owls undertake flights in silence as they
allocation stage is to optimize UEs ' QoE by simultaneously
pass towards the prey. Therefore, they obtain modified
taking into account the inter- and intra-slice priorities. As in
frequency, obeying the sound intensity inverse square law. You
Equation 10, this problem can be formulated.
s
can obtain the shift in frequency for ith owl as follows:
 us
  ru   Ii
Maximize  
  s
  I ci   Random noise
  R max   Ri
2
s S
 u s U s  s  
(11)
(7)
Voles are present in the real world and therefore their
Subjectto   m,u s bm  B movement causes the owls to quietly change their approach
m  M s  S u s U s location. The prey movement is planned using chance and
7(a) thus new owl positions can be obtained through the following
Rs
m in
 rus  R s
m ax position updating function:

 O i   * I ci * |  V  O i |, if p vm  0 . 5 
t t

7(b) t 1
 
Oi 
In which limit (7a) specifies that the sum of sub-channels t t
 O i   * I ci * |  V  O i |, if p vm  0 . 5 
 
allocated can not exceed the maximum available bandwidth;
2584
IJSTR©2019
www.ijstr.org
 INTERNATIONAL JOURNAL OF SCIENTIFIC & TECHNOLOGY RESEARCH VOLUME 8, ISSUE 11, NOVEMBER 2019 ISSN 2277-8616

(12)
In the above equation pvm seems to be the possibility of Comparison Chart
optimal global progression i.e. data rate, α is a uniformly
distributed random number in the range [0, 0.5] and β is a

bandwidth (kbps
linearly decreasing constant from 1.9 to 0. makes major
500
changes and encourages searching space exploration [26].
0
4 RESULT AND DISCUSSIONS user 1 user2 user3 user4
This section offers an analysis of the quality of our proposed Slice
mechanism for allocating resources. To this end, we
introduced a resource allocation algorithm that allocates
Fig. 4. Bandwidth comparison for system with and without
network resources to optimize users ' overall QoE by taking optimization
into account their QoS criteria (minimum and maximum data
rate) as well as each user's priority. We use the MATLAB to
analyze various scenarios to test the proposed models in this
paper. Thus we implement the resource allocation scheme, Resource allocation before
by taking into account the admission control process, which is optimization
done as a first step before the allocation of the resources. The
explanation behind this selection is to highlight the effect of

Resource
600

demand
admission control on network resource management. We
assume in our simulations that the UEs arrival data rate is 400
distributed uniformly over the entire simulation period. The 200
total number of UEs is distributed evenly among the slices user 1 user2 user3 user4
considered. The values of UEs within the same slice are
generated randomly with the restriction of a sum equal to 1. In Fig.5. Resource allocation based on demand without
case, the priorities of UEs are much the same as our proposed optimization
solution is considered, with the exception that the restriction of
having a number of priorities equal to 1 is applied to all users
in the system. The core network parameters being used are
about 5 MHz LTE bandwidth, about 200 overall UEs, about
Resource allocation after
10s overall interval and about 174dBm / Hz overall Noise 700 optimization
Spectral Density.
Resource
demand

200
Bandwidth

500
(kbps)

-300 user 1 user2 user3 user4

0 Fig. 6. Resource allocation based on demand with

user 1user2 user3 user4 optimization

Users in a slice
In fig 4, our proposed solutions significantly increase the data
rate for all service slices in contrast to the management of
Fig. 3. Average bandwidth for UEs resource allocation without optimization. We may note that the
proposed approach ensures higher bandwidth efficiency
compared to the management approach without optimization,
Figure 3 shows the average bandwidth of the same slices for thereby emphasizing the importance of the proposed
UEs. It can be remembered that, although these services admission control in achieving better resource utilization. From
belong to two different slices with different priorities, the this estimate, as the number of UEs rises, we can note that
proposed mechanism experiences the same throughput the total QoE decreases. This is because it algorithm helps to
bandwidth. This illustrates strongly that the proposed solution increase the overall data rate of the network by allocating
is not only capable of ensuring slice-based resource resources to the services when the number of UEs increases.
allocation, but also based on utility. Figure 5 and 6 shows that before optimizing the search based
on owl, the sharing of resources differs as shown in Fig.5 and
Fig.6. The resources are fairly dependent on the requirements
after owl search-based optimization. It is therefore noted that
there is ample use of the limited resource across network
slices.

2585
IJSTR©2019
www.ijstr.org
 INTERNATIONAL JOURNAL OF SCIENTIFIC & TECHNOLOGY RESEARCH VOLUME 8, ISSUE 11, NOVEMBER 2019 ISSN 2277-8616

4 CONCLUSION industry standard for virtualization", 2013

Network slicing is a process which allows the development of [11] Y. Zaki, "Future mobile communications: LTE
several logical networks on behalf of a common physical optimization and mobile network,” Springer Science &
infrastructure shared. In this paper, we address the issue of Business Media, 2012.
sharing resources equally between slices based on demands [12] L. Zhao, M. Li, Y. Zaki, A. Timm-Giel, and C. G¨org,
and so we implement a novel resource allocation system for "LTE virtualization: From theoretical gain to practical
network slicing in an LTE network that allows us to provide solution", in Teletraffic Congress (ITC), 2011 23rd
real-time slices and self-scale slices optimally. We formulated International, San Francisco, California, pp. 71–78,
the optimal allocation of resources as a convex problem with Sep 2011.
the goal of optimizing the overall data rate function of the [13] M. Li, L. Zhao, X. Li, X. Li, Y. Zaki, A. Timm-Giel, and
system. We implemented an owl-based search approach to C. Gorg, "Investigation of network virtualization and
solve the problem of process optimization and theoretically load balancing techniques in lte networks", in 2012
demonstrated that the approach is special and also converges IEEE 75th Vehicular Technology Conference (VTC
to the optimal solution of the global system. The proposed Spring), Yokohama, Japan, pp. 1–5, May 2012.
solution is implemented in MATLAB and simulation results [14] B. Liu, and H. Tian, "A bankruptcy game-based
have been given to test the bandwidth and resource allocation resource allocation approach among virtual mobile
efficiency of the distributed scheme for different users based operators", IEEE Communications Letters, vol. 17, pp.
on demand. We also contrasted their success with our 1420–1423, July 2013.
proposed metaheuristic approach and without it. [15] H. Halabian, "Distributed Resource Allocation
Optimization in 5G Virtualized Networks", IEEE
Journal on Selected Areas in Communications, vol.
REFERENCES 37, no. 3, pp. 627-642, 2019.
[1] S. Iyer, "Virtualisation of network functions and the [16] Y. Sun, M. Peng, S. Mao, and S. Yan, "Hierarchical
SDN", Australian Journal of Telecommunications and Radio Resource Allocation for Network Slicing in Fog
the Digital Economy, vol. 2, no. 2, 2014. Radio Access Networks," IEEE Transactions on
[2] Center for Energy-efficient Telecommunications, "The Vehicular Technology, vol. 68, no. 4, pp. 3866-3881,
Power of Wireless Cloud", tech. rep., Bell Labs and 2019.
University of Melbourne, April 2012. [17] K. Teague, M. Abdel-Rahman, and A. MacKenzie,
[3] D. G. Ainscough, "The evolution and future of modern "Joint Base Station Selection and Adaptive Slicing in
mobile communications", blueadmiral.co.uk, 2001. Virtualized Wireless Networks: A Stochastic
[4] L. E. Li, Z. M. Mao, and J. Rexford, "Toward software- Optimization Framework", 2019 International
defined cellular networks", in Proceedings of the 2012 Conference on Computing, Networking and
European Workshop on Software Defined Networking, Communications (ICNC), Honolulu, HI, USA, USA,
EWSDN ’12, Washington, DC, USA, pp. 7–12, IEEE pp. 859-863, Feb 2019.
Computer Society, 2012. [18] Y. Tun, N. Tran, D. Ngo, S. Pandey, Z. Han, and C.
[5] J. Schulz-Zander, L. Suresh, N. Sarrar, A. Feldmann, Hong, "Wireless Network Slicing: Generalized Kelly
T. H¨uhn, and R. Merz, "Programmatic orchestration Mechanism-Based Resource Allocation," IEEE
of wifi networks", in Proceedings of the 2014 USENIX Journal on Selected Areas in Communications, vol.
Conference on USENIX Annual Technical 37, no. 8, pp. 1794-1807, 2019.
Conference, USENIX ATC’14, Berkeley, CA, USA, pp. [19] A. Fendt, C. Mannweiler, L. Schmelz, and B. Bauer,
347–358, USENIX Association, 2014. "A Formal Optimization Model for 5G Mobile Network
[6] A. Gudipati, D. Perry, L. E. Li, and S. Katti, "SoftRAN: Slice Resource Allocation," 2018 IEEE 9th Annual
Software defined radio access network", in Information Technology, Electronics and Mobile
Proceedings of the Second ACM SIGCOMM Communication Conference (IEMCON), Vancouver,
Workshop on Hot Topics in Software Defined BC, Canada, pp. 101-106, Nov 2018.
Networking, HotSDN ’13, New York, NY, USA, pp. 25– [20] L. Tang, Q. Tan, Y. Shi, C. Wang, and Q. Chen,
30, ACM, 2013. "Adaptive Virtual Resource Allocation in 5G Network
[7] K.-K. Yap, R. Sherwood, M. Kobayashi, T. Y. Huang, Slicing Using Constrained Markov Decision Process,"
M. Chan, N. Handigol, N. McKeown, and G. Parulkar, IEEE Access, vol. 6, pp. 61184-61195, 2018.
"Blueprint for introducing innovation into wireless [21] G. Wang, G. Feng, W. Tan, S. Qin, R. Wen, and S.
mobile networks", in Proceedings of the Second ACM Sun, "Resource Allocation for Network Slices in 5G
SIGCOMM Workshop on Virtualized Infrastructure with Network Resource Pricing," GLOBECOM 2017 -
Systems and Architectures, VISA ’10, New York, NY, 2017 IEEE Global Communications Conference,
USA, pp. 25–32, ACM, 2010. 2017.
[8] X. Jin, L. E. Li, L. Vanbever, and J. Rexford, "SoftCell: [22] J. Chen, Y. Ma, H. Kuo, C. Yang, and W. Hung,
Scalable and flexible cellular core network "Software-Defined Network Virtualization Platform for
architecture", in Proceedings of the Ninth ACM Enterprise Network Resource Management," IEEE
Conference in Emerging Networking Experiments and Transactions on Emerging Topics in Computing, vol.
Technologies, CoNEXT ’13, New York, NY, USA, pp. 4, no. 2, pp. 179-186, 2016.
163–174, ACM, 2013. [23] P. H. A. Rezende, and R. M. M. Edmundo, "An
[9] Y. Zaki, L. Zhao, C. Goerg, and A. Timm-Giel, "LTE adaptive network slicing for LTE radio access
wireless virtualization and spectrum management", in networks," 2018 Wireless Days (WD), Dubai, United
WMNC2010, Budapest, Hungary, pp. 1–6, Oct 2010. Arab Emirates, April 2018.
[10] X. Project, "The xen project, the powerful open source
2586
IJSTR©2019
www.ijstr.org
INTERNATIONAL JOURNAL OF SCIENTIFIC & TECHNOLOGY RESEARCH VOLUME 8, ISSUE 11, NOVEMBER 2019 ISSN 2277-8616

[24] M. Hu, et al, "Dynamic slicing and scheduling for

wireless network virtualization in downlink LTE
system," 2016 19th International Symposium on
Wireless Personal Multimedia Communications
(WPMC), Shenzhen, China, pp. 153-158, Nov 2016.
[25] Y. Hao, et al, "Network slicing technology in a 5g
wearable network," IEEE Communications Standards
Magazine, vol. 2.1, pp. 66-71, 2018.
[26] M. Jain, S. Maurya, A. Rani, and V. Singh, "Owl
search algorithm: A novel nature-inspired heuristic
paradigm for global optimization," Journal of
Intelligent & Fuzzy Systems, vol. 34, no. 3, pp. 1573-
1582, 2018.

2587
IJSTR©2019
www.ijstr.org