A Systematic Review of Scheduling Algorithms and Resource

Management in Context-aware Applications: A Meta-analytic Approach

Fernando Emilio Puntel

, Andrea Char

, Maria Helena Franciscatto

ao Carlos Damasceno Lima

and Cristiano Cortez da Rocha

Universidade Federal de Santa Maria, Santa Maria, Brazil

Centro de Inform

atica e Automac¸

ao do Estado de Santa Catarina (CIASC), Florian

opolis, Brazil

Keywords:

Scheduling Algorithms, Resource Management, Context-awareness, Job Scheduling.

Abstract:

Computer resource management and scheduling algorithms have been exploited in order to run applications

in an efﬁcient and effective way. Since this area is heavily exploited and has a range of application domains,

it is possible to apply various techniques from other distributed computing areas, such as context-aware appli-

cations. In this paper, we present a systematic literature review that addresses context-aware applications with

resource management and scheduling algorithms. In total, 11 studies were selected from years 2000 to 2017,

which covered the inclusion criteria of this systematic review, presenting techniques and improvements for the

area. From the analyzes of the selected studies, it was found a diversity of application domains using a variety

of technologies.

1 INTRODUCTION

A number of resource management and job schedul-

ing techniques are currently being used in various

types of high-performance applications. With com-

puting resources and the need to efﬁciently perform

job scheduling in applications, the administrator must

have some functions such as optimizing the use of re-

sources, minimizing the jobs waiting time, reducing

energy consumption, reducing the cost to complete

an application and performing resource management

(Paul and Aggarwal, 2014).

With a large number of available applications and

evaluation methods, many approaches have been pro-

posed to improve resource and job managements.

Context-aware computing is one of the areas that has

been highlighting, where systems are able to adapt to

the environment. Since this area is relatively new and

it is closely linked to distributed computing, it is nec-

essary to identify and understand contributions, au-

thors and institutions that maintain relevant publica-

tions. Without this factor, it is difﬁcult to identify do-

mains that may receive further research in the future.

In this systematic review, we present the results of

studies found in the literature that address the research

topic, in order to explore context-aware systems re-

lated to resource management and job scheduling. In

this way, we seek to present the state of the art through

a meta-analytic approach, analyzing authors, institu-

tions and other aspects.

Four digital libraries that index computing studies

were used, and papers from years 2000 to 2017 were

analyzed. After the selection by the inclusion criteria

and three stages of papers analysis, 11 studies were

selected.

The structure of the present review is presented

as follows. In the next section the research proto-

col used in this review is presented. The Section 3

presents an overview of the selected papers. The Sec-

tion 4 presents the results obtained from the search

performed through a meta-analytic approach. The

Section 5 discusses limitations of this study and in

Section 6 the ﬁnal considerations are presented.

2 RESEARCH PROTOCOL

A systematic literature review is a process based on

necessary information to identify, evaluate and select

relevant studies in the area of interest. It is a process

controlled by a research protocol deﬁned at the begin-

ning of the systematic review to provide consistency

and robustness in the results (Tranﬁeld et al., 2003).

As a ﬁrst step, a research protocol was developed

for selection of the studies in this systematic review.

664

Puntel, F., Charão, A., Franciscatto, M., Damasceno Lima, J. and Cortez da Rocha, C.

A Systematic Review of Scheduling Algorithms and Resource Management in Context-aware Applications: A Meta-analytic Approach.

DOI: 10.5220/0006799406640670

In Proceedings of the 20th International Conference on Enterprise Information Systems (ICEIS 2018), pages 664-670

ISBN: 978-989-758-298-1

In order to be selected (in addition to the inclusion cri-

teria), the published studies: (1) must that have a re-

lationship with scheduling algorithms on any context-

aware platform or (2) application with context-aware

resource management.

The search was performed in four digital libraries:

IEEE Xplore, SpringerLink, ACM Digital Library and

Scopus. The digital libraries were chosen according

to afﬁnity with the computing area. Firstly, a pri-

mary word was chosen and then, it was combined

with words related to the area. The word combina-

tions used on all platforms are shown in Table 1.

Table 1: Combinations used in the searches.

“Context awareness AND job scheduling”

“Context awareness AND resource manager”

“Context awareness AND job management system”

The databases searches were set up so that the re-

sulting studies presented at least one of the word com-

binations in title, abstract or keywords. In each digital

library it was necessary to conﬁgure the search to re-

trieve only studies related to the research area. For

searches performed in the IEEE Xplore database, the

advanced search was used to selected only conference

publications, as well as studies published in journals

and magazines. For searches performed in the ACM

Digital Library, only studies published in confer-

ences and magazines were selected. For SpringerLink

searches, studies published in conferences and jour-

nals were selected, belonging to Computer Science

and Software Engineering/Programming and Oper-

ating Systems sub-areas. For Scopus searches, only

studies in the sub-areas corresponding to Computing

were selected.

For papers selection, the studies should meet the

following pre-established inclusion criteria: (1) com-

plete study, starting from 5 pages, (2) published in

congresses or journals and available in one of the

four databases, (3) related to context-aware applica-

tions with scheduling and resource management tech-

niques, (4) published in English, (5) published be-

tween 2000 and 2017 and (6) available for access in

the CAPES portal (CAPES, 2017).

2.1 Keywords Deﬁnition

This subsection presents concepts related to knowl-

edge areas involved in this systematic review, in-

cluding context-awareness, resource manager, job

scheduling and job management system.

• Context-awareness: context can be interpreted

as “any information that can be used to character-

ize the situation of an entity (person, place or ob-

ject) that is considered relevant to the interaction

between the user and an application, including

the user and application themselves” (Dey, 2001).

Context-awareness, in turn, allows the modeling

of systems that are able to get contextual infor-

mation from circumstances in which they operate,

reacting to that context based on rules or artiﬁ-

cial intelligence. In the last years, this concept

has come to be considered part of a process that

involves the user, thus producing general and so-

phisticated contexts models (Abowd et al., 1998)

ur et al., 2016).

• Resource Manager: the main functions are re-

lated to receiving job requests for execution, al-

locating and monitoring resources. The manager

must also be smart enough to leave the running

nodes idles as short as possible (Eijkhout, 2014).

• Job Scheduling: the scheduling algorithm is re-

sponsible for receiving and responding to user re-

quests, which implies deciding when and where

the execution for each request will begin. Since

there can be multiple concurrent requests, con-

ﬂicts occur and they must be resolved through the

scheduling algorithm (Yahyapour, 2002) (Hoves-

tadt et al., 2003).

• Job Management System: the job management

system is a component in the cluster that is re-

sponsible for controlling jobs of the users. The

main goals of the job management system are to

efﬁciently use the nodes, provide a job submission

interface for users and allow the users to conﬁgure

their cluster (Eijkhout, 2014).

2.2 Data Selection and Collection

The ﬁrst step of this systematic review was the re-

moval of duplicate papers in the databases, and stud-

ies that did not meet the initial inclusion criteria. Sub-

sequently, the titles, keywords and abstract of each

study were read. According to the inclusion criteria

for this review, studies were included when they meet

these criteria in the title, keywords or abstract. Af-

ter the selection of the studies, information and ba-

sic characteristics of each one were extracted such as

title, authors, keywords, abstract and authors afﬁlia-

tion. We also analyzed locals of publications, cita-

tions of all studies, frequency of words, among other

information to build the present review.

With the searches performed, 34 articles were

found using the pre-established combinations. Of

these, 2 articles were excluded because they were du-

plicated in the databases or did not meet the inclusion

criteria, leaving 32 articles for title, keywords and ab-

stract readings. From these readings, 11 articles were

A Systematic Review of Scheduling Algorithms and Resource Management in Context-aware Applications: A Meta-analytic Approach

665

Table 2: Papers selected and general information.

Approach Selected Through Database Published in

Impact

Factor

Citations

Context-aware job scheduling

for cloud computing environments

(Assunc¸

ao et al., 2012)

Title, abstract

and keywords

IEEE Xplore

IEEE Fifth International Conference on

Utility and Cloud Computing (UCC)

- 26

CASH: context-aware scheduler for

Hadoop (Kumar et al., 2012)

Title and

abstract

ACM Digital

Library

International Conference on Advances

in Computing, Communications and

Informatics

- 28

Interaction-aware energy management

for wireless network cards

(Crk et al., 2008)

Keywords

ACM Digital

Library

ACM SIGMETRICS Performance

Evaluation Review

- 16

A context-aware approach to emergency

management systems

(Bhavanishankar et al., 2009)

Title and

abstract

ACM Digital

Library

International Conference on Wireless

Communications and Mobile Computing:

Connecting the World Wirelessly

- 9

Bridging the application knowledge

gap: using ontology-based situation

recognition to support energy-aware

resource scheduling (H

ahnel et al., 2014)

Abstract

ACM Digital

Library

13th Workshop on Adaptive and Reﬂective

Middleware

- -

On improving resource utilization and

system throughput of master slave job

scheduling in heterogeneous systems

(Hsu et al., 2008)

Title and

keywords

SpringerLink The Journal of Supercomputing 1.326 16

Performance evaluation of a discovery

and scheduling protocol for multihop

and hoc mobile grids (Gomes et al., 2009)

Title, abstract

and keywords

SpringerLink Journal of the Brazilian Computer Society

0.707

(printed)

Big media healthcare data processing

in cloud: a collaborative resource

management perspective (Das et al., 2017)

Abstract and

keywords

SpringerLink

Cluster Computing (The Journal of Net-

works, Software Tools and Applications)

2.040 1

Building mobile multimedia services:

a hybrid cloud computing

approach (Kovachev et al., 2014)

Abstract SpringerLink Multimedia Tools and Applications 1.530 26

Supporting ubiquitous IMS-based

teleconferencing through discovery

and composition of IMS and web

components (Doolin et al., 2008)

Keywords SpringerLink

Journal of Network and Systems

Management

1.588 7

Bringing context to Apache Hadoop

(Cassales et al., 2014)

Abstract and

keywords

Scopus

8th International Conference on Mobile

Ubiquitous Computing

- 3

selected because they covered the initial inclusion cri-

teria of this review. The Figure 1 illustrates the studies

selection protocol applied in the papers.

Figure 1: Stages of the study selection process.

3 SELECTED WORKS

Through the methodology used in this SMS (Section

2), 11 papers were selected. An overview of the char-

acteristics of the studies are presented in Table 2.

Assunc¸

ao et al. (2012) presented a model that

aims to rationalize the use of computational resources

in cloud environments, combining context awareness

techniques and an adaptable job scheduler. Through

the analyzes performed with social simulations, gains

were obtained in terms of performance, quality of

service and reduction of wasted jobs. Kumar et al.

(2012) proposed a new context-based scheduling al-

gorithm for Hadoop (CASH), in order to make the

scheduler aware and take advantage of the cluster’s

heterogeneity. Crk et al. (2008) proposed mech-

anisms that are aware of user interaction through

screenshots and mouse events classiﬁcation. The ap-

proach produces signiﬁcant improvements in terms

of energy savings, accuracy, punctuality and com-

putational overhead. Bhavanishankar et al. (2009)

proposed an emergency management architecture that

processes request layers with a context-aware ap-

proach. The proposal performs event management,

time management and computational resource man-

agement. H

ahnel et al. (2014) presented a generic

approach to resource management based on ontol-

ogy, where the task scheduler is context-aware and

it gets information about application execution. Ex-

periments have shown that the system can perform

situation recognition for resource management within

4 seconds. Hsu et al. (2008) presented an efﬁcient

strategy for master slave job scheduling in heteroge-

neous underlying networks. The experiments com-

pared with other algorithms presented higher through-

put and response time. Gomes et al. (2009) pre-

sented a discovery and scheduling protocol for mobile

ad-hoc networks. The experiments presented posi-

tive results in load balancing between nodes, reduc-

ICEIS 2018 - 20th International Conference on Enterprise Information Systems

666

ing runtime and maintaining acceptable levels in mo-

bility scenarios. Das et al. (2017) presented a local

and global cloud confederation model for executing

heterogeneous requests, in order to support services

that involve health data processing. Kovachev et al.

(2014) proposed i5CLoud, a hybrid cloud architec-

ture for scalable and fast mobile multimedia services

for the market. In the experiments performed, par-

ticipants evaluated the quality of context-aware ser-

vices provided by i5CLoud, completing short-term

tasks that simulated documentation in a cultural her-

itage. Doolin et al. (2008) addressed a research

approach that combines generalized computing tech-

niques with IMS network principles to facilitate com-

position of communication sessions based on the con-

text of users. Finally, Cassales et al. (2014) pre-

sented improvements to Hadoop, by introducing con-

text awareness into scheduling algorithms. Experi-

ments have demonstrated that context awareness al-

lows Hadoop to scale based on real availability of

resources, improving task allocation patterns and ra-

tionalizing the use of resources in heterogeneous dy-

namic networks.

4 RESULTS OF THE REVIEW

THROUGH A META-ANALYTIC

APPROACH

After the application of the research protocol and

the inclusion/exclusion criteria in the papers, a meta-

analytic approach was performed in the resulting

studies. The meta-analysis has taken several aspects

into consideration, which are presented in the follow-

ing subsections.

4.1 Authors Afﬁliations

First, it was analyzed the afﬁliation location of all au-

thors, illustrated in Figure 3. All afﬁliations were an-

alyzed, since several studies have authors from differ-

ent institutions and countries. For this reason, the sum

of the afﬁliations in the Figure 3 exceeds the total of

papers selected in the review. From the obtained data,

we can observe the great distribution of countries that

make researches related to the area. This may be due

to the fact that it is a new area with many questions to

be addressed.

4.2 Citations and Locals of Publication

Table 2 presents the total number of citations that each

study obtained. We can observe that even though the

studies are relatively new, the studies present a rela-

tively high citation total, which represents an area of

great scope and growth.

We can also observe in Table 2, information that

was collected on papers publication. Of the 11 se-

lected studies, 5 studies were published in journals

and there is a great diversity of publication vehicles.

It is important to note that all journals have an im-

pact factor according to their respective websites and,

considering the scope of published journals, this area

opens up several possibilities for research. Among the

journals, it is possible to observe that some of them

have a well-open scope for computing, whereas oth-

ers are of a very speciﬁc scope, such as Multimedia

Tools and Applications and Journal of Network and

Systems Management. The rest of the studies were

published in international congresses, in which we

can observe consolidation in large areas and accep-

tation of context-aware applications.

4.3 Publication Year

Figure 2 illustrates the number of studies selected per

year of publication. We can observe that the studies

were published in a maximum of 10 years, showing

that it is a relatively new research area.

Figure 2: Publication year of the studies.

4.4 Words Frequency in Abstracts

Figures 4 and 6 show the frequency of words in the

abstracts of all the selected studies. As we can ob-

serve, there are a lot of platforms such as Cloud Com-

puting and Grids, which appear more often, since

contextual searches represent a recent area of re-

search. The observed frequency is due to the fact

that cloud computing and grids are taking the place

of clusters since around 2007 and represent hot topics

due to their abilities (Wang et al., 2008).

There is a great use of words related to energy

management, computer networks, architecture and re-

sources, which represent areas with many issues to be

A Systematic Review of Scheduling Algorithms and Resource Management in Context-aware Applications: A Meta-analytic Approach

667

Figure 3: Authors afﬁliation.

answered, especially with respect to energy manage-

ment. Likewise, there are terms widely addressed in

the studies, such as ubiquitous computing, pervasive

computing, and virtualization. It is important to note

that the concept context-awareness was not taken into

account, since it is one of the primary words of the

search for this systematic review.

Figure 4: Number of occurrences of keywords.

4.5 Total of Papers by Subarea

Since this area is in great expansion, we approach

all types of resource management and job scheduling,

therefore, the selected studies address different areas.

In this way, subareas were created from the selected

studies, in which it is possible to classify the studies.

The Figure 5 illustrates the total number of studies by

main subarea and papers classiﬁed in more than one

subarea.

Figure 5: Total of papers by subarea.

5 LIMITATIONS

Although the search protocol was designed to cover

the largest number of studies related to the area, some

studies could not be included, since it was necessary

to limit the terms searched to include the databases

and the period of publication previously mentioned.

Following the research protocol and the large num-

ber of papers, in the Springer and Scopus databases

it was necessary to select only one category and one

subcategory, thus explaining possible studies missing

ICEIS 2018 - 20th International Conference on Enterprise Information Systems

668

Figure 6: Words frequency in the studies abstracts.

on these search engines.

Another critical point concerns the different ter-

minologies used in the studies. Although there is a

similarity between the terms, some studies use the

terms differently, which may lead to the exclusion of

some studies. To avoid misclassiﬁcation, we adopted

an accepted deﬁnition for the keywords used through-

out this systematic review.

Some studies may also have been excluded be-

cause of the publication date, since this review re-

trieved publications only from 2000 to 2017. Previ-

ous studies, or studies that were published since then

could meet the review criteria. Even in this period,

studies may have been excluded by title, keywords

and abstract readings for not clearly containing the

objectives.

Due to the fact that the context-aware computing

is a recent area applied to high-performance comput-

ing, it would be possible that we would not ﬁnd many

studies if the inclusion criteria were restricted, so this

systematic review addressed all methodologies for re-

source management and scheduling algorithms, inde-

pendent of the platform that was carried out in the

studies. Despite this, only 11 papers were selected to

compose this systematic review, indicating a very low

number of relevant researches that effectively covered

the areas of interest and the pre-deﬁned criteria.

6 CONCLUSIONS

The use of resource management systems and

scheduling algorithms applied to distributed systems

and applications can be performed in several ways, al-

ways taking into account the needs of each user. Due

to this area have several applications, we seek to ﬁnd

and classify studies related to context-aware comput-

ing.

This systematic literature review on context-aware

applications with resource management systems and

scheduling algorithms in systems and clusters pre-

sented different application domains. Since it is a

relatively new area of distributed platform research,

several subareas have been found and categorized in

this review.

After the primary and secondary selections, 11

complete studies were analyzed. In this context, af-

ter an analysis of the title, keywords and summary

of the selected studies, a meta-analytic approach was

presented, showing the main studies in the ﬁeld, pub-

lication locals, countries with more authors and the

most recurrent terms.

From these considerations, our future work in-

cludes analyzing and validating the experiments of

the present study, and implementing the same proto-

col in a broader area of context-aware applications.

Also, we aim to expand the qualitative analysis per-

formed and provide recommendations to the profes-

sionals that can be useful as a guide for further re-

search.

ACKNOWLEDGEMENTS

The authors would like to thank CAPES for partial

funding of this research and UFSM/FATEC through

project number 041250 - 9.07.0025 (100548).

REFERENCES

Abowd, D., Dey, A. K., Orr, R., and Brotherton, J. (1998).

Context-awareness in wearable and ubiquitous com-

puting. Virtual Reality, 3(3):200–211.

Assunc¸

ao, M. D., Netto, M. A. S., Koch, F., and Bianchi,

S. (2012). Context-aware job scheduling for cloud

computing environments. In 2012 IEEE Fifth Inter-

national Conference on Utility and Cloud Computing,

pages 255–262.

Bhavanishankar, R., Subramaniam, C., Kumar, M., and

Dugar, D. (2009). A context aware approach to emer-

gency management systems. In Proceedings of the

2009 International Conference on Wireless Commu-

nications and Mobile Computing: Connecting the

A Systematic Review of Scheduling Algorithms and Resource Management in Context-aware Applications: A Meta-analytic Approach

669

World Wirelessly, IWCMC ’09, pages 1350–1354,

New York, NY, USA. ACM.

CAPES (2017). Portal de Peri

odicos da CAPES. Dispon

ıvel

em: http://www. periodicos. capes. gov. br. Acesso

em: 08/2017.

Cassales, G., Charao, A., Kirsch-Pinheiro, M., Souveyet,

C., and Steffenel, L. (2014). Bringing context to

apache hadoop. In 8th International Conference on

Mobile Ubiquitous Computing, Rome, Italy.

Crk, I., Bi, M., and Gniady, C. (2008). Interaction-aware

energy management for wireless network cards. In

Proceedings of the 2008 ACM SIGMETRICS Inter-

national Conference on Measurement and Modeling

of Computer Systems, SIGMETRICS ’08, pages 371–

382, New York, NY, USA. ACM.

Das, A. K., Adhikary, T., Razzaque, M. A., Alrubaian, M.,

Hassan, M. M., Uddin, M. Z., and Song, B. (2017).

Big media healthcare data processing in cloud: a col-

laborative resource management perspective. Cluster

Computing, pages 1–16.

Dey, A. K. (2001). Understanding and using context. Per-

sonal and ubiquitous computing, 5(1):4–7.

Doolin, K., Mullins, R., Abad, R. M., Moreno, M. G.,

Mota, T., Farshchian, B. A., and G

omez, M. (2008).

Supporting ubiquitous ims-based teleconferencing

through discovery and composition of ims and web

components. Journal of Network and Systems Man-

agement, 16(1):92–112.

Eijkhout, V. (2014). Introduction to High Performance Sci-

entiﬁc Computing. Lulu. com.

Gomes, A. T. A., Ziviani, A., Lima, L. d. S., and Endler,

M. (2009). Performance evaluation of a discovery and

scheduling protocol for multihop ad hoc mobile grids.

Journal of the Brazilian Computer Society, 15(4):15–

29.

ahnel, M., Mendez, J., Thost, V., and Turhan, A.-Y.

(2014). Bridging the application knowledge gap: Us-

ing ontology-based situation recognition to support

energy-aware resource scheduling. In Proceedings of

the 13th Workshop on Adaptive and Reﬂective Middle-

ware, ARM ’14, pages 3:1–3:6, New York, NY, USA.

ACM.

Hovestadt, M., Kao, O., Keller, A., and Streit, A. (2003).

Scheduling in hpc resource management systems:

Queuing vs. planning. In Job Scheduling Strategies

for Parallel Processing, pages 1–20. Springer.

Hsu, C.-H., Chen, T.-L., and Park, J.-H. (2008). On im-

proving resource utilization and system throughput of

master slave job scheduling in heterogeneous systems.

The Journal of Supercomputing, 45(1):129–150.

Kovachev, D., Cao, Y., and Klamma, R. (2014). Build-

ing mobile multimedia services: a hybrid cloud com-

puting approach. Multimedia tools and applications,

70(2):977–1005.

Kumar, K. A., Konishetty, V. K., Voruganti, K., and Rao,

G. V. P. (2012). Cash: Context aware scheduler for

hadoop. In Proceedings of the International Con-

ference on Advances in Computing, Communications

and Informatics, ICACCI ’12, pages 52–61, New

York, NY, USA. ACM.

Paul, D. and Aggarwal, S. K. (2014). Multi-objective evo-

lution based dynamic job scheduler in grid. In Com-

plex, Intelligent and Software Intensive Systems (CI-

SIS), 2014 Eighth International Conference on, pages

359–366. IEEE.

Tranﬁeld, D., Denyer, D., and Smart, P. (2003). Towards a

methodology for developing evidence-informed man-

agement knowledge by means of systematic review.

British journal of management, 14(3):207–222.

Wang, L., Tao, J., Kunze, M., Castellanos, A. C., Kramer,

D., and Karl, W. (2008). Scientiﬁc cloud computing:

Early deﬁnition and experience. In 2008 10th IEEE

International Conference on High Performance Com-

puting and Communications, pages 825–830.

Yahyapour, R. (2002). Design and evaluation of job

scheduling strategies for grid computing. PhD thesis,

Universit

at Dortmund.

ur,

O., Liu, C. H., Sheng, Z., Leung, V. C., Moreno,

W., and Leung, K. K. (2016). Context-awareness for

mobile sensing: A survey and future directions. IEEE

Communications Surveys & Tutorials, 18(1):68–93.

ICEIS 2018 - 20th International Conference on Enterprise Information Systems

670