Towards enhancing resource

David C. Wyld et al. (Eds) : CCSIT, SIPP, AISC, NLP - 2015
pp. 01–11, 2015. © CS & IT-CSCP 2015 DOI : 10.5121/csit.2015.50401
TOWARDS ENHANCING RESOURCE
SCARCE CLOUDLET PERFORMANCE IN
MOBILE CLOUD COMPUTING
Md Whaiduzzaman1
, Abdullah Gani2
and Anjum Naveed3
1,2,3
Center for Mobile Cloud Computing Research (C4MCCR)
Faculty of Computer science & Information Technology,
University of Malaya, Kuala Lumpur, Malaysia
1
wzaman@ieee.org,2
abdullah@um.edu.my,3
anjumnaveed@ieee.org
ABSTRACT
In recent years, mobile devices such as smart phones, tablets empowered with tremendous
technological advancements. Augmenting the computing capability to the distant cloud help us
to envision a new computing era named as mobile cloud computing (MCC). However, distant
cloud has several limitations such as communication delay and bandwidth which brings the idea
of proximate cloud of cloudlet. Cloudlet has distinct advantages and is free from several
limitations of distant cloud. However, limited resources of cloudlet negatively impact the
cloudlet performance with the increasing number of substantial users. Hence, cloudlet is a
viable solution to augment the mobile device task to the nearest small scale cloud known as
cloudlet. However, this cloudlet resource is finite which in some point appear as resource
scarcity problem. In this paper, we analyse the cloudlet resource scarcity problem on overall
performance in the cloudlet for mobile cloud computing. In addition, for empirical analysis, we
make some definitions, assumptions and research boundaries. Moreover, we experimentally
examine the finite resource impact on cloudlet overall performance. By, empirical analysis, we
explicitly establish the research gap and present cloudlet finite resource problem in mobile
cloud computing. In this paper, we propose a Performance Enhancement Framework of
Cloudlet (PEFC) which enhances the finite resource cloudlet performance. Our aim is to
increase the cloudlet performance with this limited cloudlet resource and make the better user
experience for the cloudlet user in mobile cloud computing.
KEYWORDS
Mobile Cloud computing, Cloudlet, Resource Scarcity, Performance Enhancement.
1. INTRODUCTION
Cloudlet is a small cloud located in close vicinity to the mobile users connected through wireless
communication. Cloudlet is installed on discoverable, localized, stateless servers running one or
more virtual machines (VMs) on which mobile devices can augment resource intensive
applications offloaded for computational resources [1]. It is a set of relatively resourceful
computers that is well-connected to the Internet and is available for use by nearby mobile devices
[2] . Satyanarayanan, et al. [1] first introduced the cloudlet concept and mentioned it as a “data
center in a box”. It is self-managing, requiring little power, Internet connectivity, and access
control for setup. This simplicity of management make it feasible to use as a model of computing

2 Computer Science & Information Technology (CS & IT)
resources and to deploy on a business premises such as a coffee shop or a doctor’s office.
Internally, a cloudlet resembles a cluster of multicore computers, with internal connectivity and a
high-bandwidth wireless LAN for external access and having the virtualization capability of
cloud computing. Hence, a cloudlet can be viewed as a surrogate or proxy of the real cloud,
located as the middle tier of a three-tier hierarchy: mobile device, cloudlet, and cloud [3-5].
Mobile cloud computing liberates mobile devices from resource constraints by enabling resource-
intensive applications to leverage cloud computing. Researchers named it as a cyber-foraging
which can be realized using distant remote cloud. However, due to WAN latency, jitter,
congestion, slow data transfer resulting increased power consumption and cost for user side [6, 7].
Hence, to alleviate these problems, clouds are being taken closer to the user by cloudlet concept.
The benefits of utilizing cloudlet are the speed of service accessibility, the support of mobility,
the enhanced application performance, the elongated battery life, and the reduced roaming data
cost[8] . Cloudlet has a major important role in Mobile cloud computing for its several
distinguished advantages and features. Recently, researchers have found cloudlet as a viable
solution for mobile cloud computing. [3, 9].Cloudlet and distant cloud have the same
functionality with some differences. Cloudlet performs the tasks that are offloaded to the cloudlet
using different offloading mechanism. Cloudlet has relatively higher resources compared to the
mobile devices and effectively, task completion time is lesser on cloudlet, compared to the mobile
device. However, unlike cloud where the user OS instance is stored along with modifications
permanently, in case of cloudlet, the basic OS instances are available while user snapshots of the
customized OS instances cannot be saved because of limited storage and lesser probability of
reuse [10, 11]. Cloudlet can be situated in common public area, business center, airport, coffee
shops, shopping mall which facilitated the offloading facility to the mobile user by connecting
the mobile devices as a thin client to the cloudlet [12, 13].
There are several methods and offloading techniques introduced for application migration from
mobile device to the computational clouds. One is VM migration, in which an already executing
VM is suspended; its processor, disk, and memory state are transferred; and finally VM execution
is resumed at the destination from the exact point of suspension. For application migration,
Satyanarayanan, et al. [4] introduced a dynamic VM synthesis that enable mobile devices to
deliver a small VM overlay to the cloudlet infrastructure that already possesses the base VM from
which this overlay was derived. The infrastructure applies the overlay to the base to derive the
launch VM, which starts the suspended execution of the suspension at the exact precise point [14,
15].
To realizing the Cyber foraging using a cloudlet, the VM-based cloudlet concept has recently
evolved to component-based cloudlets consisting of a group of computing nodes (both fixed and
mobile) that are sharing resources with one another. Software components on the mobile device
can then be redeployed at runtime to other nodes in the cloudlet according to some optimization
criteria, such as the execution time, energy consumption and throughput[7, 10] . These
applications are involving multiple users interacting with each other in a real-time fashion. The
resource-sharing concept of component-based cloudlets opens a promising research area for
collaborative applications which not only sharing computing resources but also share user data
such as processing results and context information [9, 15]. Cloudlet has a major important role in
mobile cloud computing for its several distinguished features as follows [11].
The mobile users get instant direct access to the cloudlet, due to the close proximity of the user
and the cloudlet which eliminates several drawbacks introduced by the communication latency,
jitter, and slow data transfer of cellular network. The conventional benefit of offloading the
computational resource intensive tasks into cloud is still preserved in cloudlet in which the mobile
device can get rid of resource starvation. Finally, since cloudlet is the near vicinity of the mobile
user, it can save money by avoiding expensive data charging in roaming situation [8, 16]. Hence,

Computer Science & Information Technology (CS & IT) 3
cloudlet can be situated in common public area, business center, airport, coffee shops, shopping
mall which facilitated the offloading facility to the mobile user by connecting the mobile devices
as a thin client to the cloudlet[13].Section 2 explains the problem analysis of resource scarce
cloudlet, Section 3 describes the PEFC framework with different components, and Section 4
presents the significance of the framework and finally Section 5 draws conclusive remarks with
our future research direction.
2. PROBLEM ANALYSIS
Cloudlet has finite resource and this is an intrinsic problem of cloudlet which has a negative
effect of its overall performance. In this study, we aim to highlight the resource scarcity problem
and establish the impact of the resource scarcity on cloudlet performance by empirical analysis.
2.1 System parameters
We first define the resources, tasks and workload for cloudlet. Subsequently we define the
performance parameters. Our defined several definitions for analysis are as follows:
Definition 1 (Resource): A resource represents an available unit which is required for executing
a task. We can denote r as a resource and R can be denoted as a set of available resource.
Definition 2 (Cloudlet resources): For cloudlet, three types of resources are available such as
CPU, memory and storage. The fundamental operation of most CPUs, regardless of the physical
form they take, is to execute a sequence of stored instructions.
Definition 3 (Task): A task can be a logical unit of work which is executed by resource.
Definition 4 (User service time): User service time indicate the total time taken by the cloudlet
and other related transfers time and network delay to deliver the computation service to the
mobile user. User service time indicate the total time, including execution time and wait time,
taken by the cloudlet to deliver the computation service to the mobile user.
Definition 5 (Remote application execution time): The total time is taken for execute the
application in cloudlet
Definition 6 (Local application execution Time): The time is taken to execute the program in
local mobile device.
Definition 7 (Large application/program): We purposely make this application as a compute-
intensive which comparatively take longer time than our small program.
Definition 8 (Small Application/program): We purposely make this application as a compute-
intensive which comparatively take lesser time than our large program.
2.2 Experiment
In this section, we describe experimental model, mobile client and cloud servers specification,
communication infrastructure and data design. Here, we analyze the impact and verify the
severity of resource scarcity impact on cloudlet. Considering Open stack as a cloudlet service
provider, we use mobile devices for a local user for the test bed.

One is constrained of resources of cloudlet, since these resources are free to use, therefore the
available resources are not adequate when the number of users or applications request for
computational services from the cloudlet. The increasing user load, in some point make the cloud
resource scarceness and ultimately the resource constraint of cloudlet take more time to
application execution, which affect the user experience with long application execution delay. It
also creates the problem of on demand resource availability by the new user request since
resources are provisioned already and lack of available resources for the new requests. These
effects severely hindered the main purpose of using cloudlet.
Figure 1. Cloudlet problem Analysis with possible effects in application execution time
If we consider the aforementioned problems in cloudlet which are fewer resources and less
scalable, ultimately it affects the application execution time. Among the resources in cloudlets,
for our experiment we consider the processor speed, and how effective it will be to execute the
application in cloudlet. In application execution, we consider two aspects: One is in remote
application execution time which should be in cloudlet, and other one is in local application
execution time which we consider the time taken by the local execution by the mobile device. For
our experiment, we define and design a large program and a small program to differentiate it by
two types of execution time. It is obvious that large program computation time is more than the
small program computation time. We made it purposely to done our experiment. Several
important terms for this experiment:

Figure 2. Comparison of small program execution time in cloudlet and mobile device
Figure 3. Comparison of Large Program execution time in cloudlet and mobile device
We have done this experiment in Open stack Havana Cloud running in Ubuntu12.04 Linux server
and Samsung Galaxy S2 is used as a mobile device in the lab environment real test bed.

Figure 4. Comparison of different types program execution time in cloudlet and mobile device
From Experiments and Figure 2,3,and 4 , we can confirm that all of the cases if 16 users using
the cloudlet, then 17th number user have no benefit from the cloudlet in terms of program
execution time in cloudlet. Because, in this case, the mobile device can run faster, if execute in
locally.
3. PERFORMANCE ENHANCEMENT FRAMEWORK FOR CLOUDLET
This section reports a Performance Enhancement Framework for Cloudlet (PEFC) for MCC. The
objective is to explain and address the issues of resource scarcity problems which hindered the
cloudlet performance. This section explains the architecture and operating procedure and
performance of the proposed framework.
This framework basically consists with two major building blocks. One is the user side; we
assume it a mobile device. In this study, we assume all the devices are same type in operation
perspective. In the cloud side, we call it cloudlet, a small scale cloud which is built and operated
in open source cloud software named as open stack. We propose a novel Performance
Enhancement Framework for Cloudlet (PEFC) for MCC. PEFC address the issues of resource
scariness in cloudlet by offloading and shifting some of its workload or process to the nearby
mobile device using the Wi-Fi communication and finally sent the processed output to the
specific user. To design and development of the framework, we consider several important
offloading aspects. For decision making, we employed several decision making algorithms and
we consider the mobility pattern for the framework. Two tiers architecture of our framework
basically builds up with the mobile user which we consider as an end user. This user can be
identified as a smartphone, tables and even the laptop user.
After user sending the tasks to the cloudlet, then the second part is executed by the cloudlet for
task completion. This step involves several serious consideration and efficient consecutive
process for the whole task completion cycle. First of all, the framework check the usability of the
cloudlet whether it is practically beneficial for the user to offload the task. If yes then it goes

through several sections instructed and directed by our algorithms. We explain the each steps
individually the main job and their internal execution to complete the whole task and send back
the result to the user. We present the framework components and building blocks in Fig 5. The
following section explains the components of PEFC framework:
3.1 Task Handler
Task handler is situated in Cloudlet side which first receives the task request from the mobile
device. Task handler mainly analyzes the task and makes a weight for the computational benefit
from the cloud. This task handler initially assesses the whole task and the priority and importance.
For the further analysis then it send to the aggregator and profile section.
3.2 Aggregator
The aggregator mainly aggregates all the components of the task and reorganizes and rearranges
it according to the sequence of the application. It basically incorporates the different components
of the task and reshuffles and reorganizes the random part to make it in an orderly meaningful
manner.
3.3 Profiler
Profiler mainly responsible for analyze the applications and its different components. It describes
the need of demand for computation units need to complete the task execution. In addition it can
make the all components of profile to identify and reorder the task and calculate the whole
necessary resources needed to finish the work.
Figure 5. Main building blocks of Performance Enhancement Framework for Cloudlet

3.4 Analyzer and Optimizer
After profiling, tasks are handed to the analyzer and optimizer section for further analysis based
on the profiler data. By using optimization techniques, analyzer and optimizer decide to further
processing of the request tasks which should be optimally done by the available resources for the
cloudlet and mobile devices.
3.5 Scheduler
Scheduler makes the scheduling for complete the task. It assigns the priority which devices the
task will reschedule and overall available resources for task scheduling. When task finish then it
update accordingly and make the resource free for the other sequential important task for
completion.
3.6 Partition/Decision Engine
The decision engine makes the decision of application execution. It could be the in to cloudlet or
to the mobile devices other than the sending devices. This section makes decision after analysing
the resources associated with the mobile devices resources. We employ several decision making
algorithm in this stage to get the optimized and enhanced performance by make the decision that
should be the optimized one.
3.7 Mobility Manager
Mobility manager hold the status of all the mobile devices connected to the cloudlet in the
proximity of cloudlet. It can be done by storing all the Wi-Fi signal strength registered with the
Wi-Fi zone and their signal strength confirm us the proximity of the all the nearest devices.
Among them, from the client profiler, we can have the resources latest update and the information
who are waiting for the cloudlet services by the receiving the results from the cloudlet. Mobility
is an important aspect which we also consider in our framework. From the mobility pattern
analysis, when decision engine make decision which mobile device, it is going to offloaded task
to the client devices can help the predicted mobile device.
3.8 Resource Manager
Resources indexing are keeping as a database to the cloudlet. All the available and presently used
resources are keep track by the resource manager. Therefore resource manager is responsible for
keep all the updated data and information in the cloudlet which is very important for the decision
engine to estimate the available resources from the cloudlet. Hence, every new device joining in
the network should register their resources by the profiler to the cloudlet and again, when any
device leave the network, instantly the resource manager remove the device and it’s available
resources from the database.
3.9 Migrator
Cloudlet use Migrator to transfer the portion of data or code segment or process to migrate to the
surrounded mobile devices. Obviously, decision engine using our algorithms make decision to
choose the best case mobile device. After that for sending the code, migrator is responsible of
sending to the mobile device. Migrator receives the result from the mobile device to process it
again by sending to the aggregator and other further steps.

3.10 Client Devices Resources
The resources which are embedded with the client mobile devices are mentioned or marked here
as user device resources. It could be CPU, memory or even storage. Several cases it could be the
installed software that cannot be processed by the cloudlet or clod let has not installed with the
software facility. We can have huge sensors now a day that could be great resources as well such
as GPS, camera, thermometer, location apps, and embedded other latest sensors.
3.11 Cloudlet Resources
Resources which are belongs to cloudlet are normally considered as CPU, memory, and hard
disks. In this experiment and our frame work, mainly we consider the compute-intensive
applications, hence we identified the CPU is the main resources. As, we proof in chapter 3 that
certain cases, this resources are not adequate and we call it some point it as a resource scarcity
which degrade the overall cloudlet performance negatively. We aim to consider the problem and
propose the solution to shift some application or process from the cloudlet to the nearest mobile
devices that they can act as collaborative resources for the cloudlet. Since when the loads are
reduced from the cloudlet, ultimately the performance has been increased and the results we get
from the nearest mobile devices as working as a resource node or hub eventually make the whole
process faster and make the cloudlet to increase the performance and can get rid of resource
scarcity.
4. ADVANTAGE AND SIGNIFICANT
Our proposed framework has several important and significant features and charters tics which
we outline below:
4.1 Enhanced Resources
One of the most significant features of this framework is it eventually increase the resources of
cloudlet. In fact, practically, it does not increase the physical resources of the cloudlet .However;
it does ultimately the same thing, if we increase the physical resources to the cloudlet. Especially
when it dealt with the shifting tasks from cloudlet to the mobile device, in reality, it reduces the
load from the cloudlet resources which free the cloudlet resources.
4.2 Nearest Proximity
The resource rich mobile devices are located to the close proximity of the cloudlet which brings
the mobile devices additional or extra resources nearby to the cloudlet. Moreover, this nearness
makes the framework more stable because it is less affected by the band width or network latency
which is common for distant cloud.
4.3 No Cost
In the cost perspective, no costs are involved with this model and we know the cloudlet service is
free. Hence this model is pretty straight forward no payment module involve with in it. This also
help easier to implement the model and encourage the near vicinity of the cloudlet user to use
their resources as free but still the mobile user who give his/her resources for the cloudlet ensure
the advantage of time or energy benefit. This establishes the strong feasibility of free service and
in this model the cloudlet and the mobile user both are getting mutual benefit with this model.

4.4 Performance Enhancement
Most important aspect of this framework, we get extra resources without any payment which
ultimately increase the overall performance. It works both ways, at the same time it ensure the
cloudlet performance increase and well as the mobile user performance by helping the mobile
user to perform his tasks in timely manner with time and energy benefit.
4.5 Network delay
Our proposed framework considers the resources from the cloudlet and some are from the nearest
mobile devices collaboratively. Therefore, this framework model do not need cellular or mobile
network for communication between client and cloud. Moreover, it works using the wireless
network with Wi-Fi connection, hence notably important that this proposed framework is free
from communication latency and network delay.
4.6 Adaptive ness
To make the experiment simple and to avoid complexity, we make the framework simple and use
one type of mobile devices. However, this model is suitable and appropriate for other devices
with different operating systems. In short, it is feasible and supports the heterogeneity devices and
operating system. Therefore, this framework has a strong adaptive ness with different platform
and software which make it as strong business feasibility in the marketplace for its adaptive ness.
5. CONCLUSIONS
In this paper, we highlight the cloudlet resource scarcity impact on overall performance in the
cloudlet for mobile cloud computing. First we explain several basic concepts of cloudlet, then, for
empirical analysis, we make several assumptions and research boundaries. In addition, we
experimentally examine the effects and impacts of finite resource on cloudlet overall
performance. We establish the research gap and present cloudlet finite resource problem. In this
study, we propose a framework, PEFC, to enhance the cloudlet performance. Our proposed
framework, mitigate the resources scariness of cloudlet by shifting the load to the nearest idle
mobile device to enhance the cloudlet performance. We describe the frameworks several
components and highlights it’s the important and distinct features. Finally, our framework shows
the cloudlet performance enhancement and in our future work, we will implement the framework
and empirically investigate the performance enhancement of the cloudlet in mobile cloud
computing.
ACKNOWLEDGMENTS
This work is fully funded by the Malaysian Ministry of Education under the High Impact
Research Grant of University of Malaya UM.C/625/1/HIR/MOE/FCSIT/03.
REFERENCES
[1] M.Satyanarayanan, P.Bahl, R.Caceres, and N.Davies, "The case for vm-based cloudlets in mobile
computing," Pervasive Computing, IEEE, vol. 8, pp. 14-23, 2009.
[2] A.Bahtovski and M.Gusev, "Cloudlet Challenges," Procedia Engineering, vol. 69, pp.704-711, 2014.
[3] M.R. Rahimi, "Exploiting an elastic 2-tiered cloud architecture for rich mobile applications," in
World of Wireless, Mobile and Multimedia Networks (WoWMoM), 2012 IEEE International
Symposium on a, 2012, pp. 1-2.
[4] M.Satyanarayanan, G.A.Lewis, E.J.Morris, S.Simanta, J.Boleng, and K.Ha, "The Role of Cloudlets in
Hostile Environments," IEEE Pervasive Computing, vol. 12, pp. 40-49, 2013.

[5] M.Whaiduzzaman, M.Sookhak, A.Gani, and R. Buyya, "A survey on vehicular cloud computing,"
Journal of Network and Computer Applications, vol. 40, pp. 325-344, 4// 2014.
[6] S.Bohez, T.Verbelen, P.Simoens, and B.Dhoedt, "Allocation Algorithms for Autonomous
Management of Collaborative Cloudlets."
[7] E.Ahmed, A.Akhunzada, M.Whaiduzzaman, A.Gani, S.H.Ab Hamid, and R.Buyya, "Network-centric
performance analysis of runtime application migration in mobile cloud computing," Simulation
Modelling Practice and Theory.
[8] M.Whaiduzzaman, M.N.Haque, M.Rejaul Karim Chowdhury, and A.Gani, "A Study on Strategic
Provisioning of Cloud Computing Services," The Scientific World Journal, vol. 2014, p. 16, 2014.
[9] A.Gani, G.M.Nayeem, M.Shiraz, M.Sookhak, M.Whaiduzzaman, and S.Khan, "A review on
interworking and mobility techniques for seamless connectivity in mobile cloud computing," Journal
of Network and Computer Applications, vol. 43, pp. 84-102, 8// 2014.
[10] H.Qi, M.Shiraz, A.Gani, M. Whaiduzzaman, and S.Khan, "Sierpinski triangle based data center
architecture in cloud computing," The Journal of Supercomputing, pp. 1-21, 2014/04/26 2014.
[11] M.K. Nasir and M.Whaiduzzaman, "Use of Cell Phone Density for Intelligent Transportation System
(ITS) in Bangladesh," Jahangirnagar University Journal of Information Technology, vol. 1, 2012.
[12] N.Fernando, S.W.Loke, and W. Rahayu, "Mobile cloud computing: A survey," Future Generation
Computer Systems, vol.29, pp. 84-106, 2013.
[13] M.Shiraz, M.Whaiduzzaman, and A. Gani, "A study on anatomy of smartphone," J Comput Commun
Collab, vol.1, pp. 24-31, 2013.
[14] M.Whaiduzzaman, A.Gani, N.B.Anuar, M.Shiraz, M.N.Haque, and I.T.Haque, "Cloud Service
Selection using Multi-Criteria Decision Analysis," The Scientific World Journal, 2013.
[15] M.Whaiduzzaman and A. Gani, "Measuring security for cloud service provider: A Third Party
approach," in Electrical Information and Communication Technology (EICT), 2013 International
Conference on, 2014, pp. 1-6.
[16] M.Whaiduzzaman, A.Gani, N.B.Anuar, M.Shiraz, M.N.Haque, and I.T.Haque, "Cloud Service
Selection Using Multicriteria Decision Analysis," The Scientific World Journal, vol. 2014, p. 10,
2014.

Towards enhancing resource

More Related Content

What's hot (18)

Viewers also liked (20)

Similar to Towards enhancing resource (20)

Recently uploaded (20)

Towards enhancing resource