See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/261845639

A Survey on Research in Software Reliability Engineering in the Last Decade

Conference Paper · March 2014

DOI: 10.1145/2554850.2555161

CITATIONS READS
2 639

4 authors, including:

Joicymara Xavier Rivalino Matias Jr.

Federal University of Minas Gerais Universidade Federal de Uberlândia (UFU)
6 PUBLICATIONS   7 CITATIONS    132 PUBLICATIONS   815 CITATIONS   

SEE PROFILE SEE PROFILE

Lucio Borges de Araujo

Universidade Federal de Uberlândia (UFU)
67 PUBLICATIONS   136 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Engenharia de Confiabilidade de Software: Um mapeamento sistemático da última década View project

Cloud Computing, MCC, and Data Center Infrastructure - Performance, Availability, Reliability and related issues View project

All content following this page was uploaded by Rivalino Matias Jr. on 26 April 2014.

The user has requested enhancement of the downloaded file.

 A Survey on Research in Software Reliability Engineering
in the Last Decade
†
Joicymara Xavier, †Autran Macêdo, †Rivalino Matias, ‡Lucio Borges
† ‡
School of Computer Science, School of Mathematics
Federal University of Uberlandia, Brazil
[email protected], [email protected], [email protected], [email protected]

ABSTRACT creating 7 distinct clusters for analysis, as shown in Figure 1.

In this paper, we present the findings of a systematic review of the
SRE literature, covering 300 papers in 95 publication venues
(conferences and journals) in the last twelve years. We employ
statistic techniques to help mining patterns and trends from the
analyzed data set. We identified a prevalence of purely theoretical
studies in the surveyed period, where no more than 31% are
experimental research. Also, the number of research works in
“Reliability Assessment” has clearly increased since 2002, and
especially during the last three years. On the other hand, works in
“Reliability Modeling” and “Fault Analysis” have been gradually
reduced during the last decade.
Figure 1. Papers clustered according to keywords.
Categories and Subject Descriptors
D.2.4 [Software Engineering]: Software/Program Verification – We named each cluster according to its most frequent keyword.
reliability. G.3 [Mathematics of Computing]: Probability and The resulting names (and corresponding colors) are the following:
Statistics – reliability and life testing. red = Stochastic Modeling (SM); yellow = Software Reliability
Growth Model (RGM); blue = Reliability Modeling (RM);
General Terms purple = Software Testing (STT); green = Reliability Assessment
Reliability. (RA); orange = Software Maintenance (SMN); violet = Fault
Analysis (FA).
Keywords
Software reliability, systematic review, trends. 2. RESULT ANALYSIS
The 300 selected papers covered 304 authors from 92 different
1. INTRODUCTION affiliations spread into 15 countries. Figure 2 shows the
The software complexity has grown dramatically over the past distribution of papers across the clusters.
few decades, and this tendency will certainly continue [1]. To deal
with this challenging scenario, the discipline of software
reliability engineering (SRE) [1] has continually evolved its body
of theoretical and practical knowledge. Consequently, many
research works have been developed, considering different
methodological approaches, techniques, and applications. Due to
the importance of this research area, as well as its fast growth in
the recent years, it is natural that prospective-related questions be
raised, such as: Which SRE topics have been more recently
investigated? What research approaches have been adopted in
SRE in the last decade? Where the most cited SRE works come Figure 2. Distribution of papers per clusters (KG).
from and which problems they solve? We observe a pattern where every two or three years there is an
In order to answer these questions we searched for SRE papers increase on the number of papers (i.e., 2006, 2008, and 2011). We
using the sentence “software reliability” in the following hypothesize that this occurs due to the necessary time authors
repositories: ACM Digital Library [2], IEEE Xplore [3], Science need to maturate their research works, which seems to be two
Direct [4], and Google Scholar [5], in the period from 2002 to years in average.
2013. After analyzing every paper, we selected 300 papers more Table 1 presents the affiliation of the top ten most productive
closely related to the SRE area. Later, we applied the Ward authors, in terms of number of papers in the samples analyzed.
method [6] to evaluate the similarities among the selected papers Each line of this table corresponds to one author’s affiliation; the
third column, NPP, indicates the number of the author’s published
Draft version. papers in the surveyed period (2002-2013); the fourth column
The final version of this manuscript is published in the ACM contains the average number of citations to the author’s papers.
Symposium on Applied Computing (ACM SAC 2014), March 2014.
Table 2 presents the affiliation of the top ten most cited authors in
our data set, where the fourth column, NC, corresponds to the sum
 up of the citations to the author’s papers (column NPP) according 100%
Reliability Growth Model
90%
to Google Scholar [5]. 80% Software Testing
70%
Table 1. Affiliations of the top ten most productive authors 60% Fault Analysis
50%
Affiliation Country NPP Average 40%
Software Maintenance

Hiroshima University JPN 56 6.80 30% Reliability Modeling

Duke University USA 49 34.04 20%

10%
Reliability Assessment

National Tsing Hua University TWN 48 22.02 0% Stochastic Modeling

Hiroshima University JPN 34 6.79

Chinese Univ. of Hong Kong HKG 27 33.90
University of Connecticut USA 25 25.70 Figure 3: Evolution of annual publications per cluster topic.
Tottori University JPN 25 9.08 The top three most cited experimental related papers are the
Indian Inst. of Tech. of Kanpur IND 18 15.44 following: [7] - a theoretical and experimental work with 141
North Carolina State University USA 18 46.83 citations; [8] - a purely experimental work with 115 citations; and
Beihang University CHN 16 2.00 [9] - another purely experimental work with 104 citations.
Similarly, the most frequently cited theoretical related papers are:
Table 2. The affiliation of the top ten most cited authors [10] – with 284 citations; [11] with 282 citations; and [12] with
140 citations.
Affiliation Country NPP NC
Duke University USA 49 1668 3. REFERENCES
National Tsing Hua University TWN 48 1057 [1] Lyu, M. R. Software Reliability Engineering: A Roadmap.
Chinese Univ. of Hong Kong HKG 27 916 In Future of Software Engineering. IEEE Computer Society,
North Carolina State University USA 18 843 Washington, DC, USA, 2007, 153-170.
University of Connecticut USA 25 643
[2] ACM. ACM Digital Library. ACM, http://dl.acm.org.
Microsoft Research USA 4 436
Duke University USA 5 436 [3] IEEE. IEEE Xplore - Home. IEEE,
National Tsing Hua University TWN 10 423 http://ieeexplore.ieee.org/ Xplore/home.jsp.
Hiroshima University JPN 56 381
Duke University USA 3 297 [4] Elsevier. Science Direct. Elsevier B. V. http://www.
sciencedirect.com.
We can conclude that researchers from Duke University (USA)
and National Tsing Hua University (TWN) had important [5] Google. Google Scholar. Google, http://scholar.google.com.
influence on this research field, considering both number of
published papers and citations. Indeed, the most cited paper of [6] Johnson, R. A. and Wichern, D. W. Applied Multivariate
each cluster comes from these both countries, where USA relates Statistical Analysis, Upper Sandler River, New Jersey, 2007.
to clusters SM, RA, RM, SMN, FA, and STT, and Taiwan to [7] Nagappan, N., Murphy, B. and Basili, V. R. The influence of
RGM. Additionally, we notice that one author from Duke has organizational structure on software quality: an empirical
three papers among the most cited. case study. In Proc. of the Int'l Conference on Software
In terms of research approach, we considered experimental and Engineering. ACM, NY, USA, 2008, 521-530.
theoretical works, as well as both approaches applied together. [8] Li, Z., Tan, L., Lu, S., Zhou, Y. and Zhai, C. Have things
We observe that 31% (93 out 300 papers) relates to experimental changed now?: an empirical study of bug characteristics in
works, where only 13.6% are purely experimental. We modern open source software. Workshop on Architectural
hypothesize that this low number is due to two reasons: (1) public and System Support for Improving Software Dependability.
experimental data sets in SRE are very limited; (2) producing ACM, NY, USA, 2006, 25-33.
software reliability data through experiments typically require
long time, which sometimes is unattractive for the nowadays- [9] Saff, D. and Ernst, M. D. An experimental evaluation of
scientific publication process. The clusters “Software Testing” continuous testing during development. In Proc. of the 2004
and “Reliability Modeling” bring together the majority of ACM Int'l Symp. on Software Testing and Analysis. ACM,
experimental works. On the opposite, the cluster “Software NY, USA, 2004, 76-85.
Maintenance” contains the lesser number of experimental works.
[10] Nicol, D. M., Sanders, W. H. and Trivedi, K. S. Model-
In Figure 3, we can observe the annual publication per clusters. It based evaluation: from dependability to security. IEEE
is possible to see that researches in “Reliability Assessment” have Transactions on Dependable and Secure Computing, 1,1
made significant progress since 2002, and especially during the (Oct. 2004), 48-65.
last three years. Improvements can also be observed in “Software
Testing” research works. [11] Huang, C.-Y., Lyu, M. R.-T. and Kuo, S.-Y. A unified
scheme of some Nonhomogenous Poisson process models
for software reliability estimation. IEEE Transactions on
Software Engineering, 29, 3 (March 2003), 261-269.
[12] Nagappan, N. and Ball, T. Use of relative code churn
measures to predict system defect density. In Proc. of the
Int'l Conf. on Software Eng. USA, 2005, 284 - 292.

View publication stats