Design and Implementation of Virtualization

Cloud Computing System Intelligent Terminal

Application Layer
Lavanya C
Dept Of Computerscience
Research scholar
Kpr Arts Science And Research, Coimbatore, Barathiyar University.
Mail id: lavanyacmdr@gmail.com
Dr R Ramesh
Associate professor
Dept Of Computerscience
Kpr Arts Science And Research, Coimbatore, Barathiyar University.
Mail id: proframeshr@gmail

Abstract
Cloud task scheduling has gained popularity, and mathematical models of various cloud task
scheduling scenarios can be used to optimize the drawbacks of conventional scheduling
techniques. This research suggests an enhanced krill swarm optimization technique based on
adaptive weight to enhance the virtualization data processing effect of the intelligent terminal
application layer. The task scheduling algorithm's accuracy and convergence are increased by
optimizing the task average response time ratio and cluster load balancing. Additionally, tests are
conducted in this work to confirm the efficacy of the suggested model using the CloudSim
simulation tool. Furthermore, this study suggests an application-based virtualization technique
that turns the host computer's application programs into virtualization software within the virtual
system. in order for the virtual machine to have access to it. Lastly, this study uses tests to
confirm the validity of the suggested approach, offering a theoretical framework for the
development of an intelligent terminal application layer virtualization cloud computing system.
The creation and deployment of applications are greatly facilitated by the greater flexibility,
efficiency, and security of virtual machine hardware when compared to the conventional method
of employing physical hardware.
Keywords: cloud computing, virtualization, application layer, and intelligent terminal.

Introduction
Since the inception of the concept of cloud computing, it has garnered the interest of both
domestic and international research institutions due to its virtualization, high scalability, and on-
demand service characteristics. At the same time, the development of mobile communication
technology and the popularity of mobile terminals like smartphones and tablets have also
propelled cloud computing services to the pinnacle of the times, driving demand for high-
performance cloud computing capabilities and convenient and dependable cloud service models.
First, the mainstream application model of the Internet has gradually evolved from the
Client/Server model to the Browser/Server model, which significantly reduces the development
and operation costs, improves the distribution, and can provide fast and convenient services for
cross-platform and cross- terminal users at the same time. With the gradual commercialization
of the concept of cloud computing, cloud service providers integrate their resources to assist
users acquire cloud computing services such as application landing, service landing and security
guarantee in the form of cloud computing platform [1].
With the emergence of cloud computing platform services, ordinary peo- ple can enjoy hundreds
of millions of super computing power per second, which has greatly changed the lifestyle of the
public, and subverted the traditional technology and business model, making the scale of the
Internet expand rapidly and the amount of data explode. To deal with such massive computing
Cloud service providers are continuously extending the size of physical resource pools and
adding more infrastructure resources in response to demand and the rise of different new
resource demands. On the other hand, data centers' resource consumption rate has steadily
declined, while traditional resource management techniques have progressively lagged behind
. In light of this pressing situation, how can one prudently and efficiently distribute

Both industry and academics are now very concerned about network resources, and resource
management has progressively emerged as one of the main problems with cloud computing.
Sensors, cameras, car electronics, and other embedded devices with constrained hardware
resources are examples of edge cloud computing terminal devices, which differ from central
cloud computing. Furthermore, it has little processing power, the The edge cloud is situated in a
complex and varied physical environment, and frequently the space, temperature, and power
system are insufficient to ensure the normal state. On the other hand, the edge side demands very
high computational and real-time performance. The disagreement between the two makes it
impossible for the standard CPU architecture to suit the needs of edge cloud [3].
This research suggests an enhanced krill swarm optimization algorithm based on adaptive
weights to increase the intelligent terminal application layer's virtualized data processing
efficiency. Use two goal formulas to maximize task scheduling algorithms' accuracy and
convergence: Task average instruction response time ratio and cluster load balancing A program
for virtualization that makes the internal applications of the host It is suggested to integrate a
host into virtual machines, allowing virtual machines to utilize the application-based
virtualization technique. Experiments confirm the method's dependability, offering a theoretical
framework for the development of next intelligent terminal application layer virtualization cloud
computing systems. Utilizing hardware in the form of virtual machines is more versatile,
efficient, and secure than utilizing traditional physical hardware, and it greatly facilitates
application development and deployment.
The history and importance of cloud computing, together with the use of deep learning
techniques, are what make this paper innovative. It suggests an enhanced particle swarm and
krill swarm hybrid method that is appropriate for cloud computing work scheduling, as well as
an adjustable inertia weight krill swarm algorithm. Making use of hardware in the form of virtual
machines is more adaptable, effective, and safe, making application development and
deployment much more convenient.

Task Scheduling Algorithm for Cloud Computing

Cloud computing is built on virtualization technology, which abstracts and unifies the computer
hardware resources (such as memory, networks, storage, and CPU) and software environment
(such as applications, file systems, and operating systems). Geographical boundaries can be
removed and unified administration and utilization of computer resources across many areas can
be achieved through abstraction. Multiple virtual machines running distinct operating systems
may coexist on a single physical computer thanks to virtualization technology, which also
isolates and separates the operating environments of each virtual machine.
The natural heuristic algorithm is a type of adaptable technique with high global optimization
capabilities that has a lot of promise for resolving challenging optimization issues. An
optimization system based on natural heuristics is the Krill Swarm algorithm. On how krill
colonies forage, which has been successfully used in several sectors. However, there are still
certain issues with the Krill Swarm method's parameter setup and local search capability, which
makes it difficult to use the algorithm for cloud computing tasks issues with scheduling. A novel
adaptive inertial weight-based krill swarm method (AIWKH) is proposed in this study to address
these issues. Following an iteration, the particle's inertial force weight will be reset to zero if its
fitness value drops. Additionally, the accuracy of the cluster web service and the task average
command response speed both boost the convergence of the task scheduling algorithm.

Improved Krill Swarm Algorithm Strategy

Based on the foraging habits of krill populations, the krill swarm method is a natural heuristic
optimization algorithm with a high rate of convergence and strong global optimization
capabilities. The krill population's foraging behavior, information transmission systems, and
dynamic updating tactics are the basic tenets of the algorithm.
As a versatile and globally optimized approach, natural heuristic algorithms have demonstrated
significant promise in resolving challenging optimization issues. Two naturally occurring
heuristic optimization algorithms based on group behavior—the particle swarm optimization
algorithm and the krill swarm optimization algorithm—have produced noteworthy outcomes in
their respective domains. These two algorithms do, however, still have certain limitations with
regard to parameter sets, global optimization, and local search capabilities. To completely This
research suggests a cloud computing work scheduling technique based on an enhanced particle
swarm and krill swarm hybrid algorithm that combines the benefits of these two methods.
The performance of the krill swarm algorithm is significantly influenced by its primary
parameters. To increase the algorithm's performance after addressing the optimization problem,
it's critical to use sensible baseline values. Among them, the method's step size scaling factor Ct
is chosen using the linear descent technique; the higher the Ct in the early iteration stage, the
more viable areas the program may investigate.

Ct = Ct max — 1 – t/t max Ct min

The highest and smallest step scaling factors among them are Ctmax and Ctmin, respectively.
If the value of fitness If the fitness value of the i-th particle lowers after an iteration, then its
inertia weight will be reset to zero at the next iteration, that is, wn = 0, wf = 0. If the fitness value
of the i-th particle becomes better after the iteration, then its

Figure 1 Flowchart of improved krill swarm algorithm.

At the following iteration, the inertia weight won't change.

 The step size scaling factor and the fitness function's value, f, are two of them. When
comparing the Ct determined by formula (4) to that determined by formula (1), the difference is
linear. Consequently, discrete system computation can mitigate the declining trend of Ct and
accelerate particle convergence in the following

Simulation Experiments
Experimental Environment

Simulation tools may be used to establish a repeatable and controllable computer environment in
the cloud computing operating environment, allowing users to do effective testing in that setting.
The CloudSim program offers a framework for simulating computer systems without taking into
account the specifics of associated services and infrastructure. Performance bottlenecks may be
identified beforehand with the use of simulation tools, and issues can be analyzed beforehand to
make targeted improvements. As seen in Figure 8, the SimJave layer serves as the computation
engine for discrete event simulation and is in charge of a number of essential software features,
such as scheduling and queue formation.

Cloud architecture
 intercomponent communication and the formation of cloud system entities. Cloud providers
primarily develop and implement their scheduling strategies at the CloudSim simulation layer in
order to examine the operational effectiveness of various scheduling techniques. This layer
allows the modeling of cloud-based data centers. Users can specify matching settings in the user
code layer based on the requirements of the design.

Two-layer mapping is supported by CloudSim, which is the carrier of mapping between virtual
machines and physical hosts. Figure 9 displays the CloudSim program in its operational state.
The primary purpose of CIS and DataCenter is to facilitate interactive information exchange,
while the primary role of the VMScheduler class is to manage virtual machines. It is possible to
redesign the DatacenterBroker class's methods, that is, overloaded, after which simulation testing
may confirm the benefits and drawbacks of different scheduling techniques.

Conclusion
Cloud task scheduling has gained popularity, and alternative mathematical models of cloud task
scheduling scenarios can be used to optimize the drawbacks of traditional scheduling algorithms.
Although krill swarm optimization is only suitable for solving some simple practical
optimization problems, it can still be improved and applied. This study has a deeper
understanding of the model structure of krill swarm algorithm, and also has some improvements
to the algorithm. Moreover, this paper proposes application virtualization, and it is an
application-based virtualization method that virtualizes the application programs inside the host
machine into the virtual machine and the virtual- ization software makes the virtual machine
accessible. In addition, this paper constructs the scenario modeling of cloud task scheduling
environment with experiments, and formulates the goals of optimization. The performance of
the algorithm in real-world applications may be more accurately assessed by setting realistic
models and optimization goals.
Although this article use virtual machine online migration technology to accomplish memory
balancing across several hosts, virtual machine online migration may not operate if there are
network outages or hard disk damage between hosts. Thus, the next stage will be to investigate
the development of a host memory recycling technology among several hosts that is comparable
to multi-virtual machine balloon technology. More memory resource scheduling options will be
possible by consistently allocating and using the hosts' idle memory recycling.

References
[1] Shukur, H., Zeebaree, S., Zebari, R., Zeebaree, D., Ahmed, O., and Salih, A. (2020).
Cloud computing virtualization of resources allocation for distributed systems. Journal of
Applied Science and Technology Trends, 1(2), 98–105.