Вы находитесь на странице: 1из 14

A Research Review on Energy

Consumption of Different Frameworks


in Mobile Cloud Computing

Ramasubbareddy Somula and R. Sasikala

Abstract The mobile cloud computing (MCC) is an emerging technology, and its
popularity is increasing drastically day by day. Mobile device has been constrained
from low battery power, processing capabilities, and limited storage capacity, and
MCC is facing several security issues. Users are expecting more computing power
and security from mobile devices, in order to support user, the mobile computing
integrate with cloud computing (CC) to form a MCC. Computation offloading
improves computing features of smartphones (battery power, storage, processing
capabilities) as well as user experience with device. In this paper, our main focus is
to analyze energy consumption by executing on mobile device or remote cloud, and
offloading method and level of partitioning are implemented by exploring different
parameters based on frameworks. We summarize comparison between different
energy offloading techniques.

Keywords Cloud computing (CC) ⋅ Mobile cloud computing (MCC)


Networking ⋅
Energy consumption ⋅ Offloading⋅ Algorithm
Partitioning

1 Introduction

Cloud computing has given freedom to IT companies to focus on their business


growth and developments but not just data centers and their maintenance [1–3]. Cloud
computing provides computing resources as services to user as on demand, quick
delivery, pay as usage to improve the experience of users. Cloud service provides
three main services such as platform-as-a-service (PaaS), infrastructure-as-a-service
(IaaS), and software-as-a-service (SaaS). In recent times, user preferences are

R. Somula (✉) ⋅ R. Sasikala


Computer Science and Engineering, VIT University, Vellore, India
e-mail: svramasubbareddy1219@gmail.com
R. Sasikala
e-mail: sasikal.ra@vit.ac.in

© Springer Nature Singapore Pte Ltd. 2019 129


H. S. Saini et al. (eds.), Innovations in Computer Science and Engineering,
Lecture Notes in Networks and Systems 32,
https://doi.org/10.1007/978-981-10-8201-6_15
130 R. Somula and R. Sasikala

changing in computation because new technologies and developments are growing in


mobile computing. Both mobile computing (MC) and cloud computing (CC) integrate
to invent new technology called mobile cloud computing (MCC) which will increase
all capabilities of mobile device to improve user experience with smart devices.
According to recent survey by ABI research, around 240 million enterprises are using
cloud services through portable devices by 2015 and this will increase income of
mobile cloud computing to $5.2 billion [4]. The utilization of mobile devices is
growing in different domains such as gaming, e-learning, information management
system, entertainment, health care. Even though predictions say those mobile devices
will play important role in the distant future, mobile applications will be restricted with
battery power, storage capacity, processing potential of mobile devices. Though
modern mobiles contain sufficient resources such as battery power, processing
capability, storage capacity, this is enough to perform computation-intensive
offloading tasks such as image reorganization, language processing. Mobile device
computation power comparing to servers, desktops, laptops is very low because
computation tasks put heavy loads in battery power [5–7]. Recent research and study
on cloud computing aim to enhance computation capacity and capabilities of mobile
devices by allowing client to utilize available cloud resources, software, platform, and
computing services [8, 9]. For instance, Amazon Web Services provides simple
storage services to protect user personal data [10].
The rest of the paper describes as follows. Section 2, describes related work to
our topic of interest. Section 3 describes about working process of offloading in
MCC. In Sect. 4, the comparison among different offloading frameworks has been
described. Section 5 discusses analysis of each offloading framework. Section 6
represents tree structure and discusses tools and results which are used in different
frameworks. In Sect. 7, conclusion along with feature work has been provided.

2 Background

Cloud computing
Cloud computing is an emerging technology and new way of providing com-
putation resources to users in order to access anytime from anywhere [11]. It
provides main advantages to users and business enterprises such as
(1) Offering recourses on demand basis,
(2) Pay as use,
(3) Management capabilities [12].
Cloud computing provides variety of application-as-a-service through Internet.
Cloud computing provides different clouds based on user requirement—if user
wants to store his data in public domain, cloud computing provides public cloud,
otherwise the information needs to be secured. CC provides private cloud to
required user.
A Research Review on Energy Consumption of Different Frameworks … 131

For instance, [8] AWS (Amazon Web Services) which is public computing tool
provides users two models: infrastructure-as-a-service (Iaas) and software-as-
a-service (Saas). These two models permit users to use available resources in
remote cloud [10]. Computational cloud implements different service models that
use them in different cloud environments [13].
Mobile cloud computing
In simple way to define MCC, it is a path between mobile users and cloud and
that provides enough computing resources to mobile device and processing
requirement of computational tasks on mobile devices [4]. MCC can execute and
store application out of mobile to enhance mobile computing capacity in terms of
battery energy, processing potential, storage capacity. Many business enterprises
pay attention to MCC to acquire business growth and reduce execution and
implementation cost of mobile application.
Figure 1 represents basic architecture of MCC which contains three entities as
follows:
(1) Smart devices which offload intensive computation application into cloud,
(2) Network which is required to send offload data into remote server for execution
and get result through network,
(3) Remote server (cloud) which will execute all computation-intensive applica-
tions in required remote server.
Computation offloading is process of offloading (or) sending computation-
intensive application to remote server for executing and then returns result to
mobile device. In this paper, various offloading frameworks and their approaches
are discussed below which have been published in different times. Mobile appli-
cations are partitioned to different components. Each of these components will be
executed in remote server based on comparison of different parameters. But still
several challenges of offloading need to be addressed.

Fig. 1 Computation offloading generic view of MCC


132 R. Somula and R. Sasikala

Fig. 2 Basic offloading processes

The potential of computational offloading depends on different mobile network


technologies such as Wi-fi, cellular network. They determine capability of mobile
offloading. Unlike previous Wi-fi technology, current technology is able to provide
high-bandwidth connection. Considerable amount of energy is utilized from mobile
device for data transmission using cellular network [14]. In Fig. 2, the mobile
device decides offload method B to remote server or powerful machine. Here,
remote server provides virtual resources to run offloaded components [15]. Pow-
erful machine could be a cluster or server in computing center. Offloading is useful
only when local execution consumes more energy and time than remote server. For
offloading decision (or) offloading process, many factors are involved. They are
mobile specifications, server specifications, network specifications, application
characteristics, user requirements. The resource-intensive applications such as video
games and voice reorganization are increasing day by day, so the researchers have
been trying to fill gap between user preferences and limited mobile resources by
bringing power of cloud computing to mobile device.

3 Overview of Offloading

Offloading is a method of transferring resource-intensive application from portable


device to remote server by considering different parameters. Offloading mecha-
nisms involve three tasks before it gets executed. They are partitioning [16, 17],
profiling, offloading decision.
A Research Review on Energy Consumption of Different Frameworks … 133

3.1 Application Partitioning

This is the first step of offloading mechanism which divided the entire applications
into offloadable and non-offloadable components in which components will execute
locally and which will execute in remote server in cloud, respectively. The decision
can be taken by considering multiple factors. Application developers annotate few
components for remote execution, for example for API as offloadable.

3.1.1 Profiling

This step will be helpful for users to know whether to execute locally or remotely. It
will collect all information related to network and available servers in cloud centers.
Proxy server receives and executes task at remote server on behalf of mobile
devices. Both code and data might be transferred for preparing remote execution at
cloud.

3.1.2 Decision

This is the final step of application offloading process before offloaded component
execution is started. Taking decision at runtime, more accurate information is
available. For instance, mobile might not have wireless connection or energy
consumption is too high for executing at remote server. Whenever situation
changes, the offloading can be adapted. Taking decision at runtime induces over-
head compared to taking at designing time.

3.2 Offloading Framework

There have been many frameworks for offloading intensive computation of mobile
application to remote cloud for execution. Based on that, we can classify frame-
works into two categories:
a. Virtual machine cloning
b. Code offloading
134 R. Somula and R. Sasikala

4 Comparisons Among Different Offloading Frameworks


in MCC

In this section, we will compare various offloading frameworks in three different


levels of offloading as discussed in previous section. At the end of this discussion,
we define different frameworks with respect to their own properties.

4.1 Clone Cloud

Chun et al. in [18] presented this framework which aims to reduce energy con-
sumption by offloading intensive computation components of mobile application to
remote server in cloud. In every offloading mechanism, three stages of executions
are involved as discussed in the previous section. Portioning step in this framework
with static program analysis which to find set of computations while meeting set of
constraints, for instance component that use mobile resources that should be exe-
cuted locally. The role of program analysis is to find possible migration points. In
profiling level, it will collect all information related to mobile specification, remote
cloud specification, network bandwidth, etc., based on that it will build cost model
as well as it will capture and store mobile image in cloud server. In offloading,
decision will be taken at runtime and threads are suspended in mobile device and
migrated to remote server along with states of threads; after that, threads resume on
server in order to perform computation. Both states of clone cloud and mobile
device must be synchronized periodically. After execution at server, the results will
be integrated from clone to mobile device.

4.2 MAUI

MAUI [19] is another framework which aims to reduce energy on mobile devices
for offloading process. MAUI framework is highly dynamic and hides the com-
plexity of execution at remote server and gives the impression as if execution is
done at locally. In this framework, partition is done based on developer annotations
in order to decide which components can be executed locally or which cannot. It
should meet two requirements in preparation step one is application binaries should
be installed in both mobile and server side. Proxies, profilers, and solvers should be
available in both the mobile and server side. The offloading decision is taken at
runtime by considering profiling information as an input to the solver.
A Research Review on Energy Consumption of Different Frameworks … 135

4.3 Cloudlet

Offloading computation task is not always good solution for reducing energy
consumption and improving performance of mobile device. Network latencies also
play an important role in mobile cloud computing because applications are
restricted with real-time scenario. Making cloud to available very near to cloud as
cloudlets can reduce latency rapidly. Satyanarayanan et al. in [20] proposed
framework called cloudlet that can be defined as hosting environment for offloaded
task that is deployed to remote servers. In preparation step, it offloads entire mobile
application; for that, it requires clone of mobile application processing environment
in remote server. For offloading, it would use VM as offloading mechanism and
more precisely called as dynamic VM. Cloudlet is distributed across Internet
infrastructure whose storage and computation power can be exploited by nearby
mobile users [21]. Offloading decision is taken at runtime and then offloads com-
putation task to nearby cloudlet. There is scenario like where user must execute
intensive application. At runtime, it will find nearby cloudlet for execution of
resource-intensive application [22]. Figure 3 illustrates the basic cloudlet process.

4.4 Jade

Qian et al. in [23] proposed framework that monitors both application and mobile
device status and decides right destination in order to execute computation task.
The objective of this framework is to improve energy-aware computation offloading
while reducing burden on developers for developing such application. Partition can

Fig. 3 Basic process of cloudlet


136 R. Somula and R. Sasikala

be done at class level in jade framework depends on collecting information. At


preparation, system checks energy status, workload variation, and communication
cost and developers have full control on computation components because jade
provides all APIs of programming model. Offloading decision is taken at runtime to
decide where component should be executed. Jade architecture contains two
architecture—android server and non-android servers running on operating system
such as Windows and Linux. Non-android server must have java installed because
java runtime engine runs as a java application on non-android server. Jade
dynamically changes the decision of component execution destination based on
energy consumption of mobile devices.

4.5 Mirror Server

Zhao et al. in [24] proposed mirror server framework which uses telecommunica-
tion service provider to provide voice communication services such as landline
services. Mirror server is not like other framework and extends mobile capabilities
by offering different services. In portioning, this framework uploads entire appli-
cation without any partitioning. At preparation, a new VM instance is created, and
this is called as mobile mirror. Mirror server is responsible for all mobile mirrors for
managing and deploying in computing infrastructure.

4.6 Cuckoo

Kemp et al. in [25] introduced new framework called cuckoo, and this framework
extends performance of device and reduces energy usage. This framework is
associated with java eclipse tool with android open-source framework. At partition,
cuckoo will use existing model in android to conform intensive and non-intensive
component application. In preparation stage, cuckoo requires to write offloadable
methods in both local and remote servers. Cuckoo will take decision at runtime.
Whenever remote resource is not available, then application executes locally. The
following table summarizes the different features of each framework (Table 1).
Table 1 Comparison table among different offloading frameworks
Framework Partition Profiling(or) preparing Offloading Granularity Parameters Year
decision level
VM Cloudlet The image of It requires entire mobile image which is No Whole Response time, RTT, 2009
[20] entire running available at remote server decision application unusable, annoying,
application is unacceptable
offloaded to server
Phone2Cloud The application Applications are modified manually in Static Partition or Input size, CPU 2009
[29] can be partitioned order to execute on remote server decision whole workload, delay
or offloaded application tolerance threshold,
entirely bandwidth
MAUI [19] Whether remote or Developer builds single method for twice Dynamic Method CPU consumption, 2010
local based on for both mobile and remote server. decision execution time, energy
annotation Programming reflection is used to consumption
determine whether to offload or not
Mirror server Entire application It creates a new image of phone in server No Entire 2010
[24] is offloaded decision application
Cuckoo [25] Partitioning is It is done by java VM Dynamic Method Activities, content 2010
done based on decision providers, broadcast
existing model in receivers, energy
android consumption, execution
A Research Review on Energy Consumption of Different Frameworks …

time
Clone cloud Partitioning is Duplicate of mobile device software stored Dynamic Thread Input size, device 2011
[18] done based on on server decision execution time, clone
profiling and static execution time
program analysis
137
138 R. Somula and R. Sasikala

5 Analysis

Clone cloud
Benefits
• Clone cloud improves the performance of application by offloading the correct
portion of application in a computational cloud and finds what to run and where
to run. Portioning of application and dynamic profiling can be done automati-
cally and seamlessly. Possible migration points will be identified using static
analysis, and cost model is constructed by using dynamic analysis for both
execution and migration. Offline partitioning can be done without relying on
source code, programmer annotation, but uses static analysis. Profile will con-
struct cost model by collecting data. Optimize for network, CPU speeds, and
energy consumption vs. execution time.
Issues
• Offload native code cannot be done. Local threads can read existing objects or
create new objects, but may get stalled waiting for the migrated thread. Con-
tinuous synchronization is required for shared data between the cloud and
mobile device so that communication overhead and decreased power of the
cloud are disadvantages of this paper.
MAUI
Benefits
• MAUI’s goal is to reduce the energy problems for smart devices, the mobile
industry’s primary task, and augments the energy exploitation of computational-
intensive applications; for example, face recognition, by nearly one order of
magnitude, allows latency-sensitive applications such as games, to more than
double their screen refresh rates. MAUI allows applications to bypass limita-
tions of mobile devices. Developer annotates each method with “remote” or
“local”. In MAUI, execution can be done by using net framework, but imple-
ments state migration, not code migration. Information related to device, net-
work behavior, and program can be collected by dynamic profiling. Profiling
information is given to a solver to find out the accurate code partition.
Issues
• Continuous profiling is expensive, and development effort to use MAUI requires
program annotation by the developer.
CUCKOO
Benefits
• This application model opts well for computation offloading provided by
android. Android application components are Activities, Services, Content
Providers, and Broadcast Receivers. Cuckoo mainly concentrates on activities
and services. It can be easy to utilize and understand. The programming model
A Research Review on Energy Consumption of Different Frameworks … 139

is used as interface between system and developer. By using the android’s


existing ‘activity/service’ model that separates the services (computation-
intensive parts) and activities (interactive parts of the application).
Issues
• Callbacks, security, and stateless service are not supported by cuckoo.
Cloudlet
Benefits
• Cloudlet intends to address resource shortage of mobile device by using close by
resource cloudlet. The end-to-end response time of applications executing in the
cloudlet needs to be fast (few milliseconds) and predictable. Cloudlets are
widely distributed Internet structure and decentralized. A cloudlet only main-
tains soft state, for example cache copies of data or code that is accessible
somewhere else. Cloudlet permits mobile user to exploit near mobile resources
without acquiring WAN delays and jitters. Simplifying cloudlet management is
main challenge in cloudlet. Solution is transient customization of cloudlet
infrastructure using hardware virtual machine (VM) technology. Pre-use cus-
tomization and Post-use cleanup. VM execution is migrated to destination by
suspending already executing VM process and memory state.
Issues
• WAN delay, congestion, and failures need to be addressed in cloudlet yet.
Phone2Cloud:
Benefits
• Phone2Cloud is used to reduce energy usage and application execution time on
smart devices. It consists of several components such as bandwidth monitor,
resource monitor, execution time predictor, offloading decision engine, and
offloading proxy. Phone2Cloud considers only CPU workload and input size
and predicts application execution time by using naïve technique.

6 A Tree Structure of Key Points in All Frameworks

In Fig. 4, flowchart for comparison of different frameworks and experimental


results and platform, application which have been part of individual experiments.
This diagram illustrates energy consumption in both cloud and smart device. In
clone cloud, three applications have been experimented with different sizes to
evaluate energy consumption in different environments. Application are “virus
scanning, image search, behavior profiling”, respective sizes “10 mb, 100 img,
140 R. Somula and R. Sasikala

depth5” are inputs, and the output has been shown in different execution envi-
ronments (cloud and mobile). The experiment is done under “Wi-fi, 3G” envi-
ronment. Experiment has shown better results compared to local execution.
In MAUI [19], the comparison of three different applications’ energy consumption
on the smartphone versus using MAUI for remote execution to servers. Applica-
tions are “face recognition, video game, and chess game. The experiment is done in
android environment by using both “Wi-fi and 3G” environments. The experiments
have shown that offloading is the good at achieving energy consumption. In
cloudlet, by sending requirements like timing units (T), wireless communication
performance (P), energy consumption (E), mobility routes (M) in order to find
nearest cloud server to process this service request [26], this cloudlet simulation is
done using DECM-SIM tool under cellular network. Cloudlet is an efficient one to
find nearest cloud server based on input parameters. In cuckoo, two real-time
applications have been tested in cuckoo framework, an object reorganization
application and distributed augmented reality mobile game [27], and have shown
that little work was required to enable computation offloading through cuckoo
framework. Cuckoo integrates with android and eclipse development tool under
Wi-fi environment. In Phone2Cloud, to predict execution time of application, only
take input size and CPU workload into account [28]. For experiment, they have
considered three different applications with different sizes and the output has
enhanced user experience with mobile.

Fig. 4 A tree representation of key elements in frameworks


A Research Review on Energy Consumption of Different Frameworks … 141

7 Conclusion

This paper mainly discussed cloud computing, mobile cloud computing, compu-
tation offloads. More specially, it presents distinguishing frameworks for energy
saving to improve mobile capabilities by using available resources of mobile cloud.
This paper shows different approaches that are used by several frameworks to
achieve offloading. Some of offloading can be used as static while others could be
dynamic. Even though there are many approaches for energy saving, all of them
have same concern regarding energy saving, reducing response time, and mini-
mizing the execution cost of mobile device to improve mobile capabilities. By
introducing lightweight model, that will help reducing difficulties and minimizing
effort for building, deploying, managing offloading approaches.
Feature Enhancement
Mobile cloud computing (MCC) is able to save energy, improve application, and
experience of the users. All frameworks mentioned above have their own benefits
and issues but still not up to the level to address all issues related to security,
energy, and user experience. Security issues are key problem in MCC, and they
need to be focused more compared to other issues.

References

1. R. Barga, D. Gannon, and D. Reed, “The client and the cloud: Democratizing research
computing,” IEEE Internet Comput., vol. 15, no. 1, pp. 72–75, 2011.
2. A. Huth and J. Cebula, “The basics of cloud computing,” United States Comput., 2011.
3. K. Kumar and Y.-H. Lu, “Cloud computing for mobile users: Can offloading computation
save energy?,” Computer (Long. Beach. Calif)., vol. 43, no. 4, pp. 51–56, 2010.
4. H. T. Dinh, C. Lee, D. Niyato, and P. Wang, “A survey of mobile cloud computing:
architecture, applications, and approaches,” Wirel. Commun. Mob. Comput., vol. 13, no. 18,
pp. 1587–1611, 2013.
5. M. Shiraz, A. Gani, R. H. Khokhar, and R. Buyya, “A review on distributed application
processing frameworks in smart mobile devices for mobile cloud computing,” IEEE Commun.
Surv. Tutorials, vol. 15, no. 3, pp. 1294–1313, 2013.
6. M. Randles, D. Lamb, and A. Taleb-Bendiab, “A comparative study into distributed load
balancing algorithms for cloud computing,” in Advanced Information Networking and
Applications Workshops (WAINA), 2010 IEEE 24th International Conference on, 2010,
pp. 551–556.
7. P. V. Krishna, S. Misra, V. Saritha, D. N. Raju, and M. S. Obaidat, “An efficient learning
automata based task offloading in mobile cloud computing environments,” in Communica-
tions (ICC), 2017 IEEE International Conference on, 2017, pp. 1–6.
8. M. Tulloch, Introducing Windows Azure for IT Professionals. Microsoft Press, 2013.
9. A. Klein, C. Mannweiler, J. Schneider, and H. D. Schotten, “Access schemes for mobile cloud
computing,” in Mobile Data Management (MDM), 2010 Eleventh International Conference
on, 2010, pp. 387–392.
10. S. Mathew and J. Varia, “Overview of amazon web services,” Amaz. Whitepapers, 2014.
11. P. Mell, T. Grance, and others, “The NIST definition of cloud computing,” 2011.
142 R. Somula and R. Sasikala

12. M. Armbrust et al., “A view of cloud computing,” Commun. ACM, vol. 53, no. 4, pp. 50–58,
2010.
13. R. Buyya, C. S. Yeo, S. Venugopal, J. Broberg, and I. Brandic, “Cloud computing and
emerging IT platforms: Vision, hype, and reality for delivering computing as the 5th utility,”
Futur. Gener. Comput. Syst., vol. 25, no. 6, pp. 599–616, 2009.
14. K. Yang, S. Ou, and H.-H. Chen, “On effective offloading services for resource-constrained
mobile devices running heavier mobile internet applications,” IEEE Commun. Mag., vol. 46,
no. 1, 2008.
15. G. Huerta-Canepa and D. Lee, “A virtual cloud computing provider for mobile devices,” in
Proceedings of the 1st ACM Workshop on Mobile Cloud Computing & Services: Social
Networks and Beyond, 2010, p. 6.
16. I. Giurgiu, O. Riva, and G. Alonso, “Dynamic software deployment from clouds to mobile
devices,” in ACM/IFIP/USENIX International Conference on Distributed Systems Platforms
and Open Distributed Processing, 2012, pp. 394–414.
17. J. Li, L. Huang, Y. Zhou, S. He, and Z. Ming, “Computation Partitioning for Mobile Cloud
Computing in a Big Data Environment,” IEEE Trans. Ind. Informatics, vol. 13, no. 4,
pp. 2009–2018, 2017.
18. B.-G. Chun, S. Ihm, P. Maniatis, M. Naik, and A. Patti, “Clonecloud: elastic execution
between mobile device and cloud,” in Proceedings of the sixth conference on Computer
systems, 2011, pp. 301–314.
19. E. Cuervo et al., “MAUI: making smartphones last longer with code offload,” in Proceedings
of the 8th international conference on Mobile systems, applications, and services, 2010,
pp. 49–62.
20. M. Satyanarayanan, P. Bahl, R. Caceres, and N. Davies, “The case for vm-based cloudlets in
mobile computing,” IEEE pervasive Comput., vol. 8, no. 4, 2009.
21. R. G. Alakbarov, F. H. Pashayev, and O. R. Alakbarov, “Optimal deployment model of
cloudlets in mobile Cloud Computing,” in Cloud Computing and Big Data Analysis
(ICCCBDA), 2017 IEEE 2nd International Conference on, 2017, pp. 213–217.
22. K. Ha, G. Lewis, S. Simanta, and M. Satyanarayanan, “Cloud offload in hostile
environments,” 2011.
23. H. Qian and D. Andresen, “Jade: Reducing energy consumption of android app,” Int.
J. Networked Distrib. Comput. (IJNDC), Atl. Press, vol. 3, no. 3, pp. 150–158, 2015.
24. B. Zhao, Z. Xu, C. Chi, S. Zhu, and G. Cao, “Mirroring smartphones for good: A feasibility
study,” in International Conference on Mobile and Ubiquitous Systems: Computing,
Networking, and Services, 2010, pp. 26–38.
25. R. Kemp, N. Palmer, T. Kielmann, and H. E. Bal, “Cuckoo: A Computation Offloading
Framework for Smartphones.,” in MobiCASE, 2010, pp. 59–79.
26. E. Ahmed, A. Naveed, A. Gani, S. H. Ab Hamid, M. Imran, and M. Guizani, “Process state
synchronization for mobility support in mobile cloud computing,” in Communications (ICC),
2017 IEEE International Conference on, 2017, pp. 1–6.
27. K. Ko, Y. Son, S. Kim, and Y. Lee, “DisCO: A distributed and concurrent offloading
framework for mobile edge cloud computing,” in Ubiquitous and Future Networks (ICUFN),
2017 Ninth International Conference on, 2017, pp. 763–766.
28. X. Ma, S. Zhang, W. Li, P. Zhang, C. Lin, and X. Shen, “Cost-efficient workload scheduling
in Cloud Assisted Mobile Edge Computing,” in Quality of Service (IWQoS), 2017 IEEE/ACM
25th International Symposium on, 2017, pp. 1–10.
29. F. Xia, F. Ding, J. Li, X. Kong, L. T. Yang, and J. Ma, “Phone2Cloud: Exploiting
computation offloading for energy saving on smartphones in mobile cloud computing,” Inf.
Syst. Front., vol. 16, no. 1, pp. 95–111, 2014.

Вам также может понравиться