Volume 8, Issue 5, May – 2023 International Journal of Innovative Science and Research Technology

ISSN No:-2456-2165

A Study of Network Optimization Models for

High-Performance Networks
1 2 3
Oluwatobi Akinmerese, Oludele Awodele, Kofi Makinwa,
Computer Science, Babcock Computer Science, Babcock Microelectronics, Delft University of
University, Ilishan Remo, University, Ilishan Remo, Technology, Mekelweg,
Ogun State, Nigeria Ogun State, Nigeria Delft, The Netherlands

4 5
Shade Kuyoro, Folasade Adedeji,
Computer Science, Babcock University, Ilishan Remo, Computer Science, Babcock University, Ilishan Remo,
Ogun State, Nigeria Ogun State, Nigeria

Abstract:- An important feature of a bandwidth actuality, bandwidth optimization technologies improve the
optimization system is the adequate provision of internet effectiveness of the current network. Through the use of
services with high data rates and wide coverage. Low bandwidth optimization, data packets can be relayed. The
bandwidth causes poor internet speed, network number of internet-enabled end customers who use high-
downtime, constant network traffic congestion and performance service is always increasing, increasing the
network unavailability during peak and off-peak demands on HPN. Several investments have been made by
periods, to mention a few. Existing research on large organizations to provide a secure, accessible and
bandwidth optimization focused on bandwidth efficient internet connection to their users. Based on existing
allocation in creating different channels and traffic literatures, the current state of problems has been observed
isolation to Guarantee good Quality of Service (QoS). so far and outlined viz: Poor internet speed, Network
Despite several optimization techniques and bandwidth downtime, Constant traffic congestion, energy consumption
allocation algorithms of existing researchers, there is still of internet connectivity on the rise thereby increasing
reduced connectivity by which internet users are grossly overhead cost of management, network unavailability (peak
affected in spite of increased energy cost of network and off-peak periods). In lieu of this, there is a dire need for
devices and other infrastructure. One of the serious optimization of the network because failure to optimize
issues of optimization techniques is that of the problem network has led to business being affected, connectivity
of mixed-integer linear programming. Therefore, this being marred, increased cost and network unavailability
article gives a brief overview of a list of bandwidth issues to mention a few. However, none of the existing
optimization models deployed through previous literatures has solved bandwidth optimization technical
researchers, stating the optimal algorithms that work problems. Some of the bandwidth optimization models
with each of the models while formulating a new comprised Internet traffic data model, [14][25], client server
bandwidth optimization model that solves the problem model, [1], fluid flow model, [8] [22] and user-centric
of mixed-integer linear programming technique that was model, as considered by this article. This article evaluates
the approach adopted from existing works necessary for and analyzes existing bandwidth optimization models,
wireless networks. optimal algorithms that worked with each model, as well as
a new bandwidth optimization model that solves the
Keywords:- Bandwidth, Traffic Congestion, Bandwidth problem of previous optimization techniques. In achieving
Optimization, Bandwidth Allocation. this, four sections are presented. Section 1 discusses the
introduction, giving an overview of a list of bandwidth
I. INTRODUCTION optimization models deployed by existing research. Existing
bandwidth optimization models that were compatible with
The term "big data" was coined to describe the each of the models were covered in the section 2. Section 3
enormous amounts data applications of information science develops an optimal bandwidth optimization model that
and technology on a large scale have generated over time. solves the mixed-integer linear programming problem.
Big data is carried by High-Performing Networks (HPNs), Section 4 explains the proposed optimization model. Section
which have had the ability to connect to support numerous 5 describes the modified mathematical formula for the
separate processes. [70]. Industries increased their focus on optimization model. The conclusion is showcased in section
high-performance equipment clarifications for performance 6.
issues in various network tiers as bandwidth issues on the
operations network continued to develop [18]. When
describing a high-performance network, operator networks
are called "high performance." [4], [10]], [11], [7] & [6].
Studies have shown that it was not surprising that there is
great interest in high-performance networks [16]. In

IJISRT23MAY1631 www.ijisrt.com 2664

 Volume 8, Issue 5, May – 2023 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
II. II EXISTING BANDWIDTH OPTIMIZATION introduced as a partial solution [1]. Afterward, proxy caches
MODELS were used at the proxy level and showed success [2]. Later,
cooperative caches have been introduced such as Internet
Most researchers attest to the fact that having a clear caching protocol and Cache Array routing [3]. Mutual
understanding of what high-performance networks are and synergy was introduced as a solution as well [4]. CDN
are not made the best choice for business. The internet is technology is evolved to be one of the major border-side
one of the most crucial factors for bandwidth. A network solutions to the problem of bandwidth consumption. All the
that only consists of lower-cost providers, like many small previous solutions are application-level solutions. The idea
networks, is not high-performance. Currently, a "high- behind all these solutions is to transfer data once and
performance" network is included in the insurance coverage retrieve it several times. Such a technique requires that the
of about 16% of larger companies. This is anticipated to rise data is static and retrieved. Application-level solutions are
in the near future as these networks develop and grow. employed for real-time data, such as video streaming,
High-performance networks in computer science simply compression, and scalable video coding. Some other
refer to network equipment, infrastructure, and internet network-based solutions such as integrated services
services combined with facilities that are more capable of (interval), differentiated services (DiffServ), and resource
performing high-speed communications. By initializing reservation protocol (rsvp) used either bandwidth
every Internet Protocol (IP) address without a hitch, reservation or priority classes’ creation for the data [5],[6].
optimization also makes room for having high-quality
packet travel simultaneously from sender to receiver and Table 1 Scenarios
vice versa, such as audio or video files [13]. Scenario Mean Arrival Rate (Mbps)
Type 1 Type 2 Type 3 Type 4
Since the 20th century, people's network life has 1. 0.8 1.6 2.4 3.2
become more colorful due to the rapid advancement of 2. 3.2 2.4 1.6 0.8
computer technology, the popularity of high-performance 3. 1.6 1.6 1.6 1.6
computers, and the Internet traffic data model, which has
also undergone revolutionary changes due to cloud We use the fluid simulation in [8] that approximates
computing and big data technology [14]. The development the packet-level simulation. Lee et al. [13] proposed a
of Internet application technology also confronts significant bandwidth optimization algorithm in GPS servers with
difficulties due to the influence of such a large number of multiple service classes, in which we can minimize the total
network users and voluminous and complex traffic data [9]. bandwidth while different QoS requirements for each class
As information system performance and cluster architecture queue in a multiple queue system are satisfied. Since it is
improve, issues with network bandwidth and the processing difficult to analytically evaluate the performance for Internet
efficiency of traffic are created. How to make the system traffic with self-similar and long-range dependent
run with the highest efficiency and shortest completion time characteristics, the performance was evaluated mainly using
has become an issue that affects the overall function of the simulation. In particular, fluid simulation was performed
Internet. The lack of network bandwidth in the computer instead of packet-level simulation to reduce the complexity.
data processing center is the main factor leading to this The bandwidth optimization algorithm was based on an
problem [12],[23]. Therefore, in the big data communication exterior penalty function method, using the relation between
environment, it is crucial to figure out how to address the the allocated bandwidth vector and the performance
issue of how the massive data in the network obtained from fluid simulation. However, the bandwidth
communication channel takes up too much bandwidth, optimization using an exterior penalty function required a
leading to the traditional communication network channel long time to converge because it used many simulations to
bandwidth being unable to meet the needs of data obtain the direction vector and step size. In a fluid flow
communication, poor communication quality, low speed, model, The input rate of class i in a fluid flow model is
and so on [15]. constant in a unit time interval length, δ [22]. We take into
account an environment with four classes of queues, where
The long-distance cable capacity around the world is each input traffic has a length of 213 steps with resolution δ
much less in capacity than the amount of data transferred. = 0.1s. Table I shows the mean arrival rates of the four
Bandwidth limitation is a source of bottlenecks and delays different traffic classes in five different cases. The standard
over the network. As a consequence, QoS degradation due deviation of traffic amounts during a unit time interval with
to delay and latency is a serious issue. Really thanks to length δ = 0.1s is set to the mean amount during the same
content delivery networks (CDN) and peer-to-peer unit time, and the Hurst parameter is set to 0.85. (i = 1, . . . ,
technologies (P2P), the end-to-end or the client-server n), In order to satisfy the Quality of Service (QoS) needs for
model has been optimized to avoid passing the whole applications and to advance the transition to user-centric
network. Network and application-level solutions are network architectures, bandwidth allocation and
introduced to solve the problem of the delay of un-cached management are crucial [2]. Given that bandwidth is a
data. limited resource, artificial intelligence techniques are
progressively replacing traditional methods of bandwidth
The client-server architecture was utilized when the allocation. The development of wireless technology has
Internet first started. In such a model, the bandwidth was resulted in the emergence of applications, protocols, and
consumed very fast. Local caching on the client side was scenarios that have benefited human endeavors [20]. To

IJISRT23MAY1631 www.ijisrt.com 2665

Volume 8, Issue 5, May – 2023 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
serve a variety of applications, services, and transmissions, This research suggests a brand-new algorithm to make
scalable and dependable communication networks are the most of available bandwidth for all text-based
required. Sensor networks, machine-to-machine applications. The algorithm is used for offline SMS in
communications, the Internet of Things, millimeter-wave mobile phones as well as real-time online text chatting. In
techniques, multiple input multiple output technology, and the core network, text is handled transparently without any
many more recent advancements have helped to increase the changes. The suggested approach employs the 'A-M'
effectiveness of communication and data transmission paradigm for compression. Depending on the context of the
through wireless networks [3]. The user-centric paradigm transferred data, using such an algorithm can save anywhere
based on Quality of Experience (QoE), which is decided by between 25% and 90% of the bandwidth. Other applications
QoS given by the network, is becoming more prevalent in such as email, web browsing, etc… will also gain from
mixed networking situations [5]. applying such an algorithm. Real-time text chatting and
mobile SMS apps were both demonstrated in this article.
III. OPTIMAL ALGORITHMS OF The technique relies on a pre-defined dictionary installed on
OPTIMIZATION MODELS the client side or in a cloud close to each client, with a
maximum size of 16 K bytes [1].
An algorithm for scheduling networks is called
weighted fair queueing (WFQ). According to [17], the We take into account a bandwidth optimization issue
researchers used high bandwidth utilization while in Generalized Processor Sharing servers with numerous
continuously submitting data to the users. In this class queues in order to reduce the overall bandwidth while
publication, the researchers discussed a number of still satisfying the QoS criteria for each class queue. In our
algorithmic techniques that could reduce the routing earlier paper [13], we used a simulation-based optimization
network's execution time, energy consumption, and time technique to achieve an optimum bandwidth vector because
delays. Peer-to-peer networks were provided with good it is challenging to quantitatively determine the
network services, a flexible network between users, and an performance, such as the delay distribution, with self-similar
absence of traffic regulations. To transfer data, it expanded input traffic in a GPS server. However, the previous
its bandwidth from source to destination. Additionally, it optimization algorithm requires rather a long simulation
reduced the amount of unavailable bandwidth. The time to solve the problem by using exterior penalty function
researchers employed techniques to maximize the methods. Without using complicated computations, we
bandwidth and transmission speed while transmitting the suggest a new bandwidth optimization approach based on
data in an equal amount of time. Since the bandwidth was bandwidth ratio adjustment. The time needed to determine
consumed in real-time, the article was unable to handle the the best bandwidth allocation for GPS servers is
problem of tracking the network by applying this significantly shorter in numerical findings [8] [22].
optimization technique.
In order to meet the Quality of Service (QoS) needs
An improved information application system is for applications and support the transition to user-centric
proposed by optimizing the fine integration method of fuzzy network architectures, bandwidth allocation and
fractional ordinary differential equations and combining management are crucial. Given that bandwidth is a limited
them with software-defined networking (SDN) to address resource, artificial intelligence techniques are progressively
the issues that the network bandwidth of previous replacing traditional methods of bandwidth allocation. The
information application systems cannot guarantee the Whale Optimization Algorithm (WOA) was examined in
quality of big data transmission, resulting in low this study to provide the best possible bandwidth allocation
transmission efficiency and slow data processing, etc. First, in wireless networks. WOA is a new swarm intelligence
the Pade approximation fractional-order fuzzy differential technique that imitates humpback whales' foraging
equation fine integration method is derived. The behavior. This study allocated the bandwidth to real-time
optimization formula for the adaptive selection of the users (RTUs) and nonreal-time users while reserving
weighted parameter N and the enlarged item number q is bandwidth for future users. The simulations were
then developed using the error analysis theory. An enhanced implemented in MATLAB and the results were discussed in
information application system is created when SDN is terms of connection probability with a focus on available
used, and the improved algorithm and system are found bandwidth and the numbers of RTUs on the network. The
using performance tests and example simulations. The findings showed that the suggested WOA technique
outcomes demonstrate that the upgraded method has higher effectively optimized the bandwidth allotted to customers
numerical accuracy and computing efficiency than the and demonstrated bandwidth management of the limited
improved algorithm. Additionally, compared to the previous quantity of bandwidth [2].
system, the enhanced port data merge rate and task
completion efficiency are much higher. It demonstrates that The speed at which data flows starts to slow down as
the information application system suggested in this study the volume of traffic gets close to the network's carrying
can more effectively address the issues of poor capacity. By default, new packets will be dropped if a queue
communication speed and insufficient bandwidth in buffer on an interface reaches capacity. Switches and routers
traditional systems, and it offers a fresh viewpoint for large can use Quality of Service (QoS) to queue and service
data bandwidth optimization [14]. higher-priority traffic before lower-priority traffic and to
remove lower-priority traffic in favour of higher-priority

IJISRT23MAY1631 www.ijisrt.com 2666

Volume 8, Issue 5, May – 2023 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
traffic during times of congestion [24]. The employment of 2022]. Existing works of literature regarding bandwidth
various queuing techniques to manage which packets are monitoring and optimization systems were reviewed using
forwarded (bandwidth allocation) and which packets are the Preferred Reporting Items for Systematic Reviews and
dropped (Buffer space) was discussed in [21]. Meta-Analysis (PRISMA) method. The PRISMA approach
was chosen because it was more suited to comparing
IV. DATA FLOW-INTENSIVE MODEL bandwidth monitoring and optimization models and is most
broadly applicable across various study fields. The four
The model was implemented on a network operating stages of PRISMA are identification, screening, eligibility,
centre and lastly, the prototype developed was evaluated and inclusion.
using the OMNET++ simulation tool [Researcher’s model,

Fig 1 Data Flow Intensive Model

The main purpose of the review was to classify By existing works, The system's energy consumption,
systems that have been adopted in existing bandwidth E, was dependent on the device's energy consumption, P,
monitoring and optimization models and the types of which changed with time, t. If T was the duration for a
problems found. From the existing articles, the bandwidth particular period, then energy was provided by:
optimization issues discovered were categorized with
various metrics, approaches, and systems employed by
researchers. The articles revealed the flaws in the current
systems. Firstly, the new mathematical integration equations
based on time-series routing is deduced. A two discrete
As a result of the input data for the that are two
time-series of the input data for the bandwidth optimization
discrete time series: Anti-meridian (AM) and Prime-
based on the Anti-Meridian (AM) and Prime-Meridian (PM)
Meridian (PM) in this thesis study, a modified mathematical
formula is given as:
equation [5] was initially formulated from the existing
mathematical model for the integration of (AM)/(PM) Timer
relays)

IJISRT23MAY1631 www.ijisrt.com 2667

Volume 8, Issue 5, May – 2023 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
GENERAL MATHEMATICAL analysis of existing NO-NC switches as a result of the triggering by am/pm
Formulae FOR AM/PM TIMER RELAY timer relay.

A new mathematical equation was in turn developed

which is known as ENERGY-SAVING AND COST
REDUCTION BANDWIDTH EQUATION(ESCRBE)
Take note that
V. MODIFIED MATHEMATICAL ENERGY-
SAVING AND COST REDUCTION
BANDWIDTH EQUATION
Where R (T) is any antiderivative of R(t) (that means In Wireless networks, the activities that consume most
R’(T) = R(t)). This is the fundamental theorem of network of the energy are network bandwidth transmission and
calculus. reception. The energy consumption for transmitting or
receiving data depends on the network traffic volume and
Among them the network traffic speed from the source to the destination.
Under such considerations, the expected energy to transmit
a PRTG analyzed monitoring metrics to the network
administrator is defined in Equation (6).
For Anti-Meridian (AM) Time

Where FUX is the utilized energy for data

transmission, M is the monitoring model for the Anti-
For Prime-Meridian (PM) Time Meridian (AM)/Prime-Meridian (PM) Timer relay, Famtr is
the consumed energy for transmitting or receiving data
 Decision Variables according to the morning period between 12am till 12
U(T) = utilized energy, R(T) = Consumed energy noon, while Fpmtr is the consumed energy for transmitting or
receiving data according to the prime-time between 12noon
 Objective Function till 12 midnight the following day. εno is the coefficient of
U1 (T1) + U2 (T2) = ʃ 0T1R1(t1) dt + ʃ 0T2R2(t2) dt energy consumed in the normally-open switch, εnc is the
coefficient of energy consumed in the normally-closed
switch, and eth is the number of active normally-
open/normally closed cisco wireless routers defined in
Equation (7).
= ½ [TR(t) + 0

= ½ R(t) T

= TR(t)/2
The expected energy to receive a PRTG analyzed
Where R(t) = Total Traffic Volume (sent and monitoring metrics for the anti-meridian time/prime-
received)/ Total traffic speed (sent and received) meridian time to the network administrator is calculated in
Equations (8) & (9)
R(t) = Total amount of energy utilized in network
devices FSX(M) = MFamtr (8)

As a result of the gaps registered and observed in the FSX(M) = MFpmtr (9)
formula for existing works as applied in (i) which are as
follows, Since a normal cisco wireless router ni only transmit
data to the network hosts via the network server, the
 Existing formula concentrated on power consumption following equation can calculate its energy consumption:
of network devices while leaving out the energy
consumption rate of the wireless devices F(ni) = FUX(M) (10)
 Existing formula also focused on AC power of the
network devices only while leaving out the DC source However, the utilized energy of a network host must
of power especially when there is power outage due to include the consumed energy of destination network
any form of circumstance bandwidth from host networks, AM/PM timer relay, and
 The am/pm timer relay was not eventually integrated normally-open/normally-closed switches to the Energy-
because the previous mathematical model only applied saving and cost reduction monitoring model. Therefore, the
am/pm while leaving out the automatic switching of

IJISRT23MAY1631 www.ijisrt.com 2668

Volume 8, Issue 5, May – 2023 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
energy consumption of a network host DIj is calculated as comparison with the new bandwidth optimization model
Equation (11). was presented with a new mathematical equation that solves
the integral problem of merging the am/pm timer relay.
F(DIj)=FSX(M)Oj+(Oj+1) MFWR+FUX(M) (11)
VI. CONCLUSION
where Oj is the number of Network Hosts in Server J,
And FWR Is the Total Energy Linkage For For One-Hour The EEBMOM was developed using bandwidth
Interval Of Am/Pm Timer Relay. Under The Above optimization techniques. These techniques had five methods
Considerations, The Residual Energy Of A Network which were hardware compression, deduplication, object
Host Ni Can Be Estimated By Equation (12). caching, traffic shaping, and rectifying the forward error.
Data collection for this research based on bandwidth
monitoring using the Paessler Router Traffic Grapher
(PRTG) network monitoring software was done at a
network operating centre in South-Western Nigeria. The
While The Energy Loss Of A Network Between The data collected comprised 8,673 records of bandwidth
Source (Isp) To Destination (Clients/Consumers) DIj Is monitored metrics such as Date/Time of bandwidth
Described In Equation (13). monitoring, network Traffic volume (of end-to-end
connectivity), network traffic speed (from source to
destination and vice-versa), network downtime, and
coverage. The model was formulated using bandwidth
During the initialization stage, the prtg monitoring optimization (BO) techniques while the data collected were
software is activated to monitor initial network hosts and coupled with three algorithms: FUE-sub-channel matching
gives the analysis based on the date/time, network traffic algorithm (FSMA), Joint sub-channel and power allocations
volume, network traffic speed, network downtime and algorithm (JSPA), and integrated structure cabling system
network coverage metrics based on the consumed energy. (ISCS) algorithm. The performance of the algorithms was
Once the prtg has monitored the network hosts, the am/pm evaluated based on four BO metrics: network bandwidth
timer relays will be triggered by the monitoring sensors of dropping, network bandwidth blocking, bandwidth
the prtg via the network administrator. Furthermore, the utilization and bandwidth consumption. The optimal
normally-open/normally closed switches are then triggered algorithm was used to formulate the optimization model
by the am/pm timer relays to switch-on active cisco wireless using the PRTG monitoring tool. A new mathematical
(border gateway) routers and switch off passive cisco equation was formulated from the model for the integration
wireless (border gateway) routers till whenever the network of Anti-Meridian (AM)/ Prime-Meridian (PM) Timer relays
bandwidth of the host network is about to clog up, then this and the Normally-Open (NO)/Normally-closed (NC)
process is vice-versa. This process is called the Energy- switches for energy-efficiency to design the EEBMOM. The
saving and cost reduction control model. (ESCRCM). EEBMOM was implemented by solving an optimization
problem using the mixed-integer programming optimization
The process of selecting network hosts to build an technique. The EEBMOM was tested and evaluated using
optimal network bandwidth algorithm by the by the prtg is the OMNET++ simulation software. (OSS).
described in the testing and evaluation stage.
REFERENCES
Other components used were Anti-Meridian/Prime-
Meridian (AM/PM) Timer relay which triggered the [1]. A. Mokhtar, A. Mansour and A. Fouad, “Bandwidth
Normally-open (NO), Normally-Closed (NC) switch. The Optimization for Real-Time Online and Mobile
NO-NC switch switched active routers on and made other Applications” Free Lancing, Nile Innovations, Cairo,
routers passive. Egypt
[2]. A. Tamuno-Omie, S. Orike and P. Elechi,
The bandwidth optimization model is presented on “Bandwidth Optimization of Wireless Networks
mixed-integer programming which was used to solve the Using Artificial Intelligence Technique” Department
mathematical equation involving the integration of the of Electrical Engineering, Rivers State University,
am/pm timer relay that triggers the no/nc switches both on Port Harcourt, Rivers State Nigeria
peak and off-peak modes (12 midnight till 12 noon and also [3]. C. Jiang, H. Zhang, Y. Ren, Z. Han, K. Chen, and L.
from 12 noon till 12 midnight the following day) from Hanzo, “Machine learning paradigms for next-
PRTG data collected for this study. generation wireless networks,” in IEEE Wireless
Communications, vol. 24, no. 2, pp. 98- 105, 2017.
The various algorithms developed for the bandwidth [4]. D. Oriato, S. Girdlestone and O. Mencer. “Dataflow
optimization technique vary in accuracy when applied to Computing in Extreme Performance Conditions”, in
other optimization models from the bandwidth usage [26]. Advances in Computers, Vol. 96, pp. 105-137,
Therefore, three (3) optimization algorithms namely: FUE- Elsevier (Science Publishing Co.), Amsterdam, New
sub-channel matching algorithm (FSMA), Joint sub-channel York, NY, USA 2015 doi:
and power allocations algorithm (JSPA) and integrated https://doi.org/10.1016/bs.adcom.2014.11.002.]
structure cabling system (ISCS) algorithms were applied in

IJISRT23MAY1631 www.ijisrt.com 2669

Volume 8, Issue 5, May – 2023 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
[5]. F. Farid, S. Shahrestani, and C. Ruan, “Quality of [18]. SGRwin, “TOP ten benefits of network performance
service concerns in wireless and cellular networks,” management”, Santander (Cantabria) Madrid, 2019.
in Communications of the IBIMA, vol. 20, pp. 1 – Available [Online]: https://www.sgrwin.com/top-10-
15, 2013]. benefits-of-network-performance-management/
[6]. G. Varghese, “ Demultiplexing”, in Network [19]. S. Huang, P. Wei, and B. Hualaitu, “Bandwidth
Algorithms (An Interdisciplinary Approach to optimization of information application system under
Designing Fast Networked Devices), University of fine integral method of fuzzy fractional order
California, San Diego, 2007, ch. 8, pp. 182-196, doi: ordinary differential equations” School of
https://doi.org/10.1016/B978-012088477-3/50010-2]. Mathematics and Information Engineering,
[7]. I. Grigorik, “What every web developer should know Chongqing University of Education, Chongqing,
about Networking and browser performance”, 2013 China, July, 2020.
[8]. J. Lee, “Bandwidth Optimization Algorithm Based [20]. S. Orike and T. Alalibo, “Comparative analysis of
on Bandwidth Ratio Adjustment in Generalized computer network protocols in wireless
Processor Sharing Servers” IEEE Xplore, Korea communication technology,” in International Journal
Advanced Institute of Science and Technology Dan of Electronics Communication and Computer
Keun Sung Korea publication at: Engineering, vol. 10, no. 3, pp. 76 – 85, 2019.
https://www.researchgate.net/publication/224675686 [21]. T. Velmurugan, H. Chandra and S. Balaji,
Conference Paper · July 2006 DOI: https://doi.org ”Comparison of Queuing Disciplines for
10.1109/ICC.2006.254789 · Differentiated Services Using OPNET,” in Advances
[9]. J. Li, Y. Zhang, X. Chen and Y. Xiang, “Secure in Recent Technologies in Communication and
attribute-based data sharing for resource-limited users Computing, 2009. ARTCom, Vol. 9, no. 2, pp.744-
in cloud computing”, in Journal of Computer 746, 2009].
Security, Vol. 72 pp. 1–12, 2018. [22]. V. Paxson and F. Floyd, “Wide area traffic : The
[10]. J. Melnick, “Network devices explained”, Netwrix failure of poisson modeling,” in IEEE/ACM
Corporation, Texas, USA, 2022. Transactions on Networking, vol.3, no.3, pp.226-
[11]. J. Walrand and P. Varaiya, “Network services and 244, June, 1995.
Layered Architectures. High-Performance [23]. X. Wang, L. Yang, L. Kuang and X. Liu, “A tensor-
Communication on Networks”. 2nd Edition, ch. 2, pp. based bigdata-driven routing recommendation
39-102, 2000. Elsevier doi: https://doi.org/10.1016/ approach for heterogeneous networks”, IEEE
B978-0-08-050803-0.50007-1] Network, vol. 33, no. 1, pp. 64–69, 2019.
[12]. J. Wu, M. Dong, K. Ota, J. Li, “Big data analysis- [24]. X. Zhang and T. Wang, “Elastic and Reliable
based secure cluster management for optimized Bandwidth Reservation Based on Distributed Traffic
control plane in software defined networks”, IEEE Monitoring and Control”, in IEEE Transactions on
Transactions Network. Serv. Manage. Vol. 15 no. 1, Parallel and Distributed Systems Vol. 33, No.12,
pp. 27–38, 2018. pp. 4563 – 4580, August, 2022 doi:
[13]. M. Dirani and Z. Altman. "A cooperative https://doi.org/10.1109/TPDS.2022.3196840]
reinforcement learning approach for inter-cell [25]. Z. Ding, L. Li and Y. Hu, “A modified precise
interference coordination in OFDMA cellular integration method for transient dynamic analysis in
networks." In 8th Intl. Symposium on Modelling and structural systems with multiple damping models”, in
Optimization in Mobile, Ad Hoc and Wireless Mechanical System Signal Processing, ch. 98, pp.
Networks. IEEE, 2010. 613–633, 2018.
[14]. P. Li, J. Li, Z. Huang, C.Z. Gao, “Privacy-preserving
outsourced classification in cloud computing”, in
Journal of Cluster Computing. Vol. 21 no. 1, 277–
286, 2018.
[15]. R. Arya, S.C. Sharma, Energy optimization of energy
aware routing protocol and bandwidth assessment for
wireless sensor network, Int. J. Syst. Assurance Eng.
Manage. 9 (3) (2018) 612– 619.].
[16]. R. Grant and S. L. Oliver, “Networks and MPI for
cluster computing” in Topics in Parallel and
Distributed Computing (Introducing Concurrency in
Undergraduate Courses), ch. 6, pp. 117-153, 2015.
doi: https://doi.org/10.1016/B978-0-12-803899-
4.00006-7.
[17]. R. Reddy, V. Pavan, D. Reddy, R. Manikanta and I.
Chandra, “Bandwidth optimization using different
algorithm techniques in wireless sensor
networks”. International Journal of Advance
Research, Ideas and Innovations in Technology, Vol.
4, no. 6, pp. 187-191, 2018.

IJISRT23MAY1631 www.ijisrt.com 2670