Sustainability in Software Engineering:
A Systematic Literature Review
Birgit Penzenstadler, Veronika Bauer
Coral Calero
Xavier Franch
Technische Universität München, Germany
Universidad de Castilla - La Mancha, Spain
Universitat Politècnica de Catalunya, Spain
{penzenst|bauerv}@in.tum.de
Coral.Calero@uclm.es
franch@essi.upc.edu
Abstract—Background: Supporting sustainability in software
engineering is becoming an active area of research. We want
to contribute the first Systematic Literature Review(SLR) in
this field to aid researchers who are motivated to contribute
to that topic by providing a body of knowledge as starting
point, because we know from own experience, this search can
be tedious and time consuming.
Aim: We aim to provide an overview of different aspects of
sustainability in software engineering research with regard
to research activity, investigated topics, identified limitations,
proposed approaches, used methods, available studies, and
considered domains.
Method: The applied method is a SLR in five reliable and
commonly-used databases according to the (quasi-standard)
protocol by Kitchenham et al. [1]. We assessed the 100 first
results of each database ordered by relevance with respect to
the search query.
Results: Of 500 classified publications, we regard 96 as relevant
for our research questions. We sketch a taxonomy of their
topics and domains, and provide lists of used methods and
proposed approaches. Most of the excluded publications were
ruled out because of an unfitting usage of terms within the
search query.
Conclusions: Currently, there is little research coverage on
the different aspects of sustainability in software engineering
while other disciplines are already more active. Future work
includes extending the study by reviewing a higher number
of publications, including dedicated journal and workshop
searches, and snowballing.
I. M OTIVATION AND BACKGROUND
Sustainability is currently an omni-present term in calls
for research proposals and conference sessions (ICSE,
CAiSE, RE, etc.). However, in literature, there is no
overview of the current state of the art in supporting sustainability in software engineering research and practice.
Consequently, researchers who are motivated to contribute
to that topic (like the first author [2]) have to invest much
time in finding a basic body of knowledge through literature
research of many unrelated leads.
This paper reports on our systematic literature review with
the objective of retrieving a solid basis of knowledge1 on the
support of sustainability in software engineering. The full
protocol is available online as technical report [4].
1 One of the common motivators for SLRs named by Zhang and Babar
in [3, Tab.I].
A. Definition of Sustainability
To clarify our research objective, we define our understanding of sustainability and what we mean by sustainability and how we want to apply it to software engineering.
The most cited definition of sustainable development [5] is
to “meet the needs of the present without compromising the
ability of future generations to satisfy their own needs.” According to [5], sustainable development needs to satisfy the
requirements of the three dimensions of society, economy,
and environment. A fourth dimension, human sustainability,
is less present in the public discussion. According to [6], it
should be included, as it is the basis for the other dimensions.
All four dimensions of sustainability are further detailed on
in our SLR protocol [4].
B. Sustainability Aspects in Software Engineering
Sustainability aspects can be brought to bear both during
the development and use of software systems. We distinguish
four aspects of sustainability in SE (orthogonal to the dimensions introduced in Sec. I-A). The development process
viewpoint includes:
• Development process aspect: Sustainability in the initial
system development process (with responsible use of
ecological, human, and financial resources). This aspect
focusses on the initial conceptual and constructional
development and we distinguish it from the late phase
of actual system production for reasons of analysis.
• Maintenance process aspect: Sustainability of the software system during its maintenance period until replacement by a new system (with continuous monitoring of quality, knowledge management).
The product viewpoint encompasses the aspects of sustainability during production and usage:
• System production aspect: Sustainability of the software
system as product with respect to its use of resources
for production (using green IT principles and sustainably produced hardware components). The actual
system production happens after most of the initial
development process and considers, inter alia, mass
production aspects, logistics and factory organization
issues.
System usage aspect: Sustainability in the usage processes in the application domain triggered by the
software system as product (responsible in impact on
environment, using green business processes).
We expect these aspects to have different scales of impact,
growing from small to large in the order presented above,
so that the system usage aspect potentially has the biggest
impact (and, therefore, improvement potential). However,
this is also dependent of the system under analysis.
For our SLR, we are looking for all four aspects of
sustainability in software engineering. The aspects imply
different levels of abstraction and varied granularity, but
nevertheless we are interested in the state of research for
each of them.
•
Mahaux et al. [7] performed a preliminary search on
the DBLP Computer Science Bibliography database3 . For
articles with the prefixes “sustainab-” OR “ecolog-” OR
“environmental-” in the title, the data base returned over
3000 results (in January 2010), but filtering on important
software-related venues lead to as few as 11 results. They
propose that a systematic literature review should be conducted.
In contrast to [7], we are interested in publications from
all scientifically sound venues and journals as we see great
potential for learning from other domains. Therefore, we
did not restrict this systematic literature review to softwarerelated venues, which is the main reason why we received
more results.
II. S EARCH D ESIGN AND P ROCESS
C. A Body of Knowledge for Sustainability in SE
Our research aim for the next years is to support the
development of ICT systems for environmental sustainability
(ICT4ES) with an adequate software engineering approach
that integrates the knowledge of related disciplines that are
concerned with sustainability. For that we need to build
up on existing knowledge is SE as well as disciplines
that have been related closer to sustainability, for example,
environmental informatics.
This research aim requires accumulating a body of knowledge for various reasons: justifying the basis for future
research, learning as much as possible from other domains
related to the topic, and providing a basis for other researchers as well as students who consider learning about
and contributing to this area. One commonly accepted research method for accumulating a body of knowledge is a
study in form of a systematic literature review [3].
D. Research Questions
The overall research objective of the study is to find out
what the current state of the art in supporting sustainability
in software engineering research and practice is. This is
further detailed in the following research questions:
RQ1 How much activity was there in the last 20 years?2
RQ2 What research topics are being addressed?
RQ3 What are the limitations of current research?
RQ4 How is sustainability support performed?
RQ5 Which methods are in use?
RQ6 Are there case studies available?
RQ7 Which domains are already considered?
E. Related Work
There are systematic literature reviews on different topics
in software engineering, but so far none has been conducted
that investigates the relation between sustainability and
software engineering.
2 Our hypothesis is that most publications will be much younger, so a
time span of 20 years ensures that we include all relevant ones.
The search design and procedure follow the guidelines
in [1]. As SLR research questions we directly adopted those
enumerated in Sec. I-D. The search process for this study
is based on an automated search of the following digital
libraries:
• IEEE Digital Library http://ieeexplore.ieee.org
• ACM Digital Library http://dl.acm.org
• SpringerLink www.springerlink.com
• ScienceDirect / Scopus http://www.sciencedirect.com
• Web of science http://apps.webofknowledge.com/WOS
A. Search String
The aim for our search string is to capture all results
that relate sustainability or environmental issues with software engineering or requirements for software systems. The
reasons for searching for requirements is that in this early
development phase sustainability issues should emerge. The
search string used on all databases is:
(sustainab* OR environment* OR ecolog* OR green)
AND
(software engineering OR requirement OR software system)
Although we explicitly list keywords in our search string
that rather point to environmental sustainability, we expect
to find all dimensions of sustainability.
B. Inclusion Criteria
We chose the following inclusion criteria in order to select
the relevant publications to answer our research questions:
• Publication date between 1/1/1991 - 31/12/2011
• Requirements phase of software development process
• Explicit mentioning of software engineering
• Scientific soundness
• Relevance with respect to research questions
• Analysis of sustainability-relevant application domains
• Coverage of a SW ecosystem or SW sustainability
3 http://dblp.mpi-inf.mpg.de/dblp-mirror/index.php
C. Exclusion Criteria
•
•
“Environment” used in the sense of system environment, not nature.
“Ecosystem” used as population of interacting systems,
for example, agents.
D. Roles and Responsibilities
•
•
•
•
•
Birgit Penzenstadler (TUM, principal researcher): IEEEXplore, result classification, detailed analysis
Zolboo Ochirsukh, Elena Mircheva, Duc Tien Vu, Tuan
Duc Nguyen (TUM, student research assistants): search
on ACM, Web of Science, ScienceDirect, SpringerLink
Veronika Bauer (TUM, expert reviewer): assessment of
search result classification, review of detailed analysis
Coral Calero (UCLM, expert reviewer): assessment of
search result classification and detailed analysis
Xavier Franch (UPC, expert reviewer): review of detailed analysis
E. Article Selection Process
The process was conducted as follows:
1) The researchers execute the search on each database
and save the references in bibliography files.
2) The principal researcher reads all titles and abstracts
and checks the inclusion and exclusion criteria for
each entry. Major criterion is the topic of the content.
3) The principal researcher classifies the papers and articles according to type, topic, and domain.
4) The expert reviewers reassess the classification and
inclusion/exclusion of search results. After their reassessment, we introduce an additional result classification: domain-specific papers that are interesting to
learn from but not focussed on software engineering.
5) The principal researcher extracts statistics and analyses the included results in further detail. This is
followed by a second assessment from the expert
reviewers.
F. Data Analysis
The data is tabulated to show:
• The databases and numbers of query results. (RQ1)
• Listed by database for included publications:
– Author, reference, date (RQ1)
– Publication type and type of content (RQ6)
– Topic of content (RQ2, RQ4, RQ5)
– Application domain (RQ7)
– Benefit for our body of knowledge (RQ4)
• The number of relevant publications per year. (RQ1)
• The respective venues and journals. (RQ1)
Furthermore, the findings for RQ3, RQ4 and RQ5 are
reported on separately. Due to the limitation of space, we
provide the full list of references of the primary study as
online appendix [8].
III. R ESULTS
The overall number of results for each data base is listed
in Tab. I.
Table I
N UMBER OF R ESULTS PER DATABASE
Database
IEEE Xplore
ACM Digital Library
SpringerLink
ScienceDirect / Scopus
Web of Science
Date
27/12/11
26/12/11
29/12/11
29/12/11
29/12/11
Results
319.601
104.217
500.004
10.749
80.503
All results were ordered “by relevance” as displayed by
the databases. From these results, we considered the first 100
results of each data base in our first iteration of the study.
In total, we reviewed 500 publications.
The following abbreviations are used to categorize the
results in Tab. II-VI:
• Publication: Kind of publication, e.g., Journal Article
(A), Conference Paper (CP), Workshop Paper (WP),
Book Chapter (BC), Letter to the editor (L)
• Type: Kind of content presented in the publication, e.g.,
method, experience report, empirical study, tool
• Topic: Short hint on principal content and keywords of
the paper or article
• Domain: Application or technology domain considered
in the publication, e.g. transport, aviation, embedded
systems, information systems, human aspects
• Benefit: Classification of why we consider this publication to be relevant with respect to the research
questions: Sustainability in software engineering (S in
SE), sustainability-related application domains (S App
Dom), sustainability (modeling) concept (S Concept),
sustainable software solutions (S SW Sol), sustainable
hardware solutions (S HW Sol)
RQ1: How much activity was there in the last 20 years?
We summarized the number of relevant publications per
database in Tab. VII, per year in Tab. VIII, and per publication type in Tab. IX. In the last two years, there was a
significant increase in the number of publications, and there
was no publication included that was older than 2005, so our
hypothesis for RQ1 holds. None of the results we included
are older than 2005, but we did have older search results
in the query evaluations, so this is not due to restricted
availability online. Although we executed the search queries
in late December, we already found journal articles dated to
January 2012 in the results which we included as they were
fully available.
While the venues were relatively distributed, there was an
accumulation of publications from “Environmental Modeling & Software” as well as the “Journal of Cleaner Production”. The fact that we classified many of the publications
as “software solutions” or “sustainability-related application
Table II
I NCLUDED R ESULTS FROM IEEE X PLORE , FULL R EFERENCES IN A PPENDIX [8]
Author and reference
Date
Pub.
Type
Topic
Domain
Benefit
Kung et al.
Middendorf et al.
Ulieru
Albertao et al.
Zhou et al.
2011
2009
2010
2010
2008
CP
CP
CP
CP
CP
method
method
method
metrics
method
green decision-making framework
envir. aspect in mechatronics design
design for resilience of networked critical infrastructures
sustainability performance of software
green remanufacturing engineering in structural machinery
building construction
mechatronics, robots
digital ecosystems
eBusiness
security tech.
S
S
S
S
S
App Dom
App Dom
App Dom
Concept
App Dom
Table III
I NCLUDED R ESULTS FROM ACM, FULL R EFERENCES IN A PPENDIX [8]
Author and reference
Date
Pub.
Type
Topic
Domain
Benefit
Mouton et al.
Geist et al.
O’Sullivan
Penzenstadler et al.
Zhongjia et al.
Audisio et al.
Miginsky et al.
Pennington et al.
Blevis et al.
Adomavicius et al.
Pereira et al.
Henriksen et al.
Dick et al.
Ticehurst et al.
Shih et al.
Cushing et al.
Cushing et al.
Kang et al.
Amsel et al.
Choucri et al.
Vicat-Blanc et al.
DesAutels et al.
Ramona
Mathevet et al.
Salski
Abidin et al.
Kase et al.
Prabhakar et al.
Umstatter
Vance
Pousman et al.
Fu et al.
2009
2009
2010
2011
2010
2009
2008
2008
2007
2007
2009
2007
2011
2007
2010
2007
2006
2008
2010
2011
2011
2011
2009
2007
2007
2010
2008
2010
2011
2007
2008
2011
A
A
CP
CP
CP
A
A
A
CP
A
A
A
CP
A
A
A
CP
A
CP
WP
BC
A
CP
A
CP
A
CP
A
A
CP
CP
CP
model
challenges
method
method
prototype
method
method
method
reflection
model
method
method
model
model
model
method
overview
method
tool
model
model
model
method
tool
method
tool
model
model
review
model
method
framework
habitat suitability models for river management
computer science challenges at exascale
data mining for biodiversity prediction in forests
teach sustainability in software engineering
design of self-propelled walking brush cutter
hazard management in two Alpine river basins
computer reconstruction of the biological networks
indirectly driven knowledge modeling in ecology
design critique as research to link sustainability and interactive technologies
technology roles & influence in an ecosystem model of technology evol.
knowledge discovery for coastal waters classification
public participation modeling in management of groundwater contamination
meta-design environments to motivate changes in energy consumption
assessing the sustainability of coastal lakes
butterfly and wetland ecology for context-aware ubiquitous learning
database design for ecologists
eco-informatics and natural resource management
GIS-based poultry litter management system for nutrient planning
tool for estimating the energy consumption of software
simulation modeling approach to evaluate renewable energy readiness
optical networks and cloud as architecture for a sust. future internet
explore the market price of “sustainable” notebooks
consolidating eco-economics through financial and fiscal instruments
role-playing game for collective awareness of wise reedbed use
fuzzy approach to ecological data analysis
tool to detect and predict urban growth pattern
sustainable informal it learning in community-based nonprofits
transfer scheme for energy harvested WSN gateways
review of the evolution of virtual fences
permanent coexistence for a linear response omnivory model
casual information visualization of printer data
urban media framework of social innovation and service design
ecological knowledge
high performance computing
forestry, data mining
education
digital manufacturing
GIS tool
ecological networks
knowledge models
design research
technology ecosystem
environmental monitoring
groundwater management
energy sustainability
environmental management
mobile learning
ecosystem info management
eco-informatics
decision support
green computing
energy simulation
cloud
life cycle analysis
eco-culture
environmental education
fuzzy systems
neural network
education
energy grids
electronics in agriculture
modeling
sustainability design
service design
S App Dom
S SW Sol
S App Dom
S in SE
S HW Sol
S App Dom
S SW Sol
S Concept
S Concept
S Concept
S SW Sol
s conc
S SW Sol
S Concept
S App Dom
S SW Sol
S Concept
S App Dom
S SW Sol
S App Dom
S App Dom
S App Dom
S Concept
S App Dom
S Concept
S App Dom
S Concept
S App Dom
S App Dom
S App Dom
S SW Sol
S Concept
Table IV
I NCLUDED R ESULTS FROM S CIENCE D IRECT, FULL R EFERENCES IN A PPENDIX [8]
Author and reference
Date
Pub.
Type
Topic
Domain
Benefit
Abdulaziz et al.
Alain et al.
Alexandrov
Ashraf et al.
Beusen et al.
Bovea et al.
Brown et al.
Cardona et al.
Cui et al.
Faith-Ell et al.
Fan et al.
Fuller et al.
Harmon et al.
Hughes et al.
Justyna et al.
Kalivarapu et al.
Kit et al.
Koormann et al.
Kubba
Liu et al.
Mei et al.
Naumann et al.
Rizzo et al.
Smith et al.
Tong et al.
Tseng et al.
Xu et al.
Yen et al.
Zhang et al.
Zhang et al.
2011
2006
2011
2012
2011
2012
2010
2011
2009
2006
2007
2006
2009
2010
2010
2008
2012
2006
2010
2011
2010
2011
2006
2010
2011
2010
2010
2011
2011
2010
A
A
L
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
BC
A
A
A
A
CP
A
A
A
A
A
A
review
method
assessment
method
tool
tool review
tool
model
method
case study
model
tool
evaluation
framework
review
framework
tool
model
strategy
study
review
model
evaluation
method
method
study
review
study
study
model
car parking requirements for sustainable transport development
modeling living systems, their diversity and their complexity
technical assessment and evaluation of environmental models
image data fusion for the remote sensing of freshwater environments
dynamic simulation and visualization software for mathematical models
taxonomy of ecodesign tools for integrating environmental requirements
software tool designed to verify ensemble forecasts of numeric variables
software package developed for dynamic simulation of water quality in rivers
management-oriented valuation for ecol. water requirements for wetlands
application of environmental requirements in Swedish road maintenance contracts
model for China’s energy requirements and CO2 emissions analysis
software package for optimizing connectivity in conservation planning
exploratory evaluation of the market case for green energy
determination of environmental water requirements for rivers
green roof performance towards management of runoff water quantity and quality
software framework for modeling of contaminant transport in groundwater
texture-based identification of urban slums in India using sensing data
modeling down-the-drain chemicals in rivers
green project requirements and strategies
energy requirements and carbon dioxide emissions of tourism industry
research progress of ecological water requirement in china
reference model for green and sustainable software and its engineering
dynamic systems-based software packages for ecological systems
green product design through product modularization using atomic theory
generating the plan of mandatory green space in urban systems
evaluating a firm’s green supply chain management
review on ecological engineering based engineering management
management’s role in adopting green purchasing standards in industry
costs and barriers of green property development in China
multi-source remote sensing data for estimating ecological water requirement
transport
agronomy systems
environmental modeling
applied geography
environmental modeling
clean production
environmental modeling
environmental modeling
nature conservation
cleaner production
energy
biodiversity protection
green energy
environmental modeling
ecological engineering
environmental modeling
applied geography
environmental modeling
architecture
energy
ecological informatics
software engineering
environmental modeling
lean manufacturing
urban development
cleaner production
management
business research
property development
environmental modeling
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
App Dom
App Dom
Concept
App Dom
SW Sol
in SE
SW Sol
SW Sol
App Dom
App Dom
App Dom
SW Sol
App Dom
App Dom
App Dom
SW Sol
SW Sol
Concept
Concept
App Dom
App Dom
in SE
SW Sol
Concept
App Dom
App Dom
Concept
Concept
App Dom
SW Sol
Table V
I NCLUDED R ESULTS FROM S PRINGER L INK , FULL R EFERENCES IN A PPENDIX [8]
Author and reference
Date
Pub.
Type
Topic
Domain
Benefit
Huang et al.
Liao et al.
Liu et al.
Du
May et al.
2009
2009
2008
2010
2006
CP
CP
A
A
CP
study
method
method
method
method
support green customers’ decision process on electronic commerce
multimedia stream format and green design concept for e-learning
wheat growth model multi-agent system
neural network control for greenhouse management
composing biological workflows through web services
web engineering
education
computing in agriculture
computing in agriculture
parallel processing
S
S
S
S
S
App Dom
Concept
SW Sol
SW Sol
SW Sol
Table VI
I NCLUDED R ESULTS FROM W EB O F S CIENCE , FULL R EFERENCES IN A PPENDIX [8]
Author and reference
Date
Pub.
Type
Topic
Domain
Benefit
Alvarez et al.
Balana et al.
Boose et al.
Bravi et al.
Ellison et al.
Greene et al.
Hall et al.
I-Wah
Jia et al.
Jin et al.
Kaduk et al.
Lundy et al.
McCabe
McIntosh et al.
Scheller et al.
Seppala et al.
Singh et al.
Tao et al.
Verweij et al.
Wang et al.
Wang et al.
Xu et al.
Zhang et al.
Zhou et al.
2011
2011
2007
2011
2006
2010
2011
2011
2011
2011
2011
2011
2006
2007
2010
2011
2011
2008
2010
2008
2009
2007
2011
2008
A
A
CP
A
A
A
A
A
A
A
A
A
A
A
A
A
A
CP
A
CP
CP
A
A
CP
method
review
method
assessment
model
method
method
analysis
case study
study
method
method
overview
method
method
assessment
method
model
perspective
method
tool
evaluation
assessment
method
decision making for treatment intensity in purifying plants
cost-effectiveness analysis of agri-environmental measures
reliable datasets for environmental models with an analytic web
life cycle assessment of a micromorph photovoltaic system
analytic webs support the synthesis of ecological data sets
decision analysis with exploration and evaluation phases
requirements for 3D vegetation structure from space
development and conditions of home-school cooperation
urban wetland planning in Beijing
integrated calculation of ecological water demand for basin systems
redicting the time of green up in temperate and boreal biomes
integrating sciences to sustain urban ecosystem services
sustainable building design in Australia
database design for ecologists including observation data
increasing the reliability of ecological models using SE techniques
greenhouse gas emissions and material flows in Finland
resource conservation technology in rice-wheat cropping system
UML-based green alignment selection decision making model
IT perspective on integrated environmental modeling
systematic research on the cost control of the green industry
web-based distributed certification system of green food
sustainability evaluation of a nature reserve project
combined biostabilization and landfill for solid waste
green remanufacturing engineering in structural machinery
wastewater
water pollution
ecological data sets
energy
ecological data sets
land management
environmental modeling
education
ecological complexity
water demand
climate change
geography
eco-architecture
ecoinformatics
software engineering
clean production
environment
intelligent computation
software engineering
cost control
env. science
environmental management
environmental management
security technology
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
domain”, some more as “sustainability concepts” and only
few as “sustainability in software engineering” implies that
there is still relatively little research published that could be
considered for building up a body of knowledge.
Table VII
I NCLUDED R ESULTS PER DATABASE
Name
IEEEXplore
ACM Digital Library
Springer Link
Science Direct
Web of Science
Total
Number of Included Results
5 out of 100
32 out of 100
5 out of 100
30 out of 100
24 out of 100
96 out of 500
Table VIII
I NCLUDED R ESULTS PER Y EAR
Year
1991 - 2005
2006
2007
2008
2009
2010
2011
2012
Total
Number of Results
0
9
12
11
11
21
29
3
96
Table IX
I NCLUDED R ESULTS PER P UBLICATION T YPE
Publication Type
Journal articles
Book chapters
Conference papers
Workshop papers
Letters to the editor
Technical reports
Total
Number of Results
65
2
27
1
1
0
96
App Dom
App Dom
SW Sol
App Dom
SW Sol
App Dom
SW Sol
Concept
App Dom
App Dom
App Dom
Concept
App Dom
SW Sol
in SE
App Dom
App Dom
in SE
in SE
Concept
SW Sol
App Dom
App Dom
App Dom
RQ2: What research topics are being addressed?
For a quick illustrated overview, we have generated a
weighted topic cloud from keywords, taken from the titles
and abstracts, that visualizes the topics in Fig. 1. We have
derived a taxonomy for the addressed research topics in
Fig. 2 that abstracts from some of the details listed in the
original classifications tables in Tab. II-VI. The dimensions
of the taxonomy are the degree of domain specifity, from
general purpose to domain-specific research and the indexing
between analytical approaches (frameworks and assessment)
and constructive approaches (methods and tools). The taxonomy shows a tendency towards domain-specific, constructive
approaches. There are not many publications rated as general
purpose, and there is little methodical guidance for supporting sustainability.
Both the keyword cloud and taxonomy rely strictly on
keywords taken from titles and abstracts. Nevertheless, their
reproduction might reveal slightly varied results, but we do
not consider that a problem as we use them only to give
an overview of topics without deriving any further statistics
from them.
RQ3: What are the limitations of current research?
To identify limitations of current research, we reviewed
our classification of topics and application domains in
Tab. II-VI. We performed a pragmatic and informal gap
analysis that resulted in three major limitations:
• High complexity.
Reason: Due to the high connectivity between the
different aspects of sustainability, (software) systems
engineering becomes highly complex. This is visible
in knowledge management approaches, e.g., [9] and
decision support systems, e.g., [10].
Conclusion: High complexity requires clear concept
definitions and consistent, traceable models. One
method to cope with high system complexity that might
prove helpful is systems’ thinking [11].
Figure 1.
Weighted topic cloud, created with http://www.wordle.net/
ecological engineering
analytical (frameworks)
green decision-making framework
green purchasing standards in industry
evaluation of environmental models
green property development in China
agri-environmental measures
framework of social innovation
evaluation of a nature reserve project
cost control of the green industry
green customersʼ decision on electronic commerce
market case for green energy
assessing the sustainability of coastal lakes
green supply chain management
sustainable building design in Australia
price of sustainable notebooks
urban ecosystem services
technology roles and paths of influence
knowledge modeling in ecology
environmental aspect in mechatronics
valuation method for ecological water requirements
micromorph photovoltaic system
architecture for a sustainable future internet
green and sustainable software and its engineering
green space in urban systems
meta-design environments to motivate changes in energy consumption
constructive (methods)
treatment intensity in purifying plants
environmental water requirements for rivers
data mining for biodiversity prediction
requirements in Swedish road maintenance
decision analysis with exploration phase
sustainability performance of software
Chinaʼs energy requirements
habitat suitability models
certification system of green food
energy requirements of tourism industry
eco-informatics and natural resource management
wetland planning in Beijing
simulation modeling approach to evaluate renewable energy green roof performance
modeling living systems
teach sustainability
modeling
in
management
of
groundwater
contamination
environmental modeling
greenhouse gas emissions and material flows in Finland
green alignment selection decision making model
green design concept for e-learning
linear response omnivory model
product modularization using atomic theory
green remanufacturing engineering
butterfly and wetland ecology for context-aware ubiquitous learning
3D vegetation structure from space
sustainable transport development
database design for ecologists
synthesis of ecological data sets
down-the-drain chemicals in rivers
network control for greenhouse management
hazard management
biostabilization and landfill for solid waste
fuzzy approach to ecological data analysis
poultry litter management system for nutrient management
energy harvested WSN gateways
conditions of home-school cooperation
design for resilience of infrastructures
green up in boreal biomes
it learning in community-based nonprofits
predict urban growth pattern awareness of wise reedbed use
environmental models with an analytic web
remote sensing of freshwater environments
reconstruction of biological networks
resource conservation technology in rice-wheat cropping
composing biological workflows
modeling of contaminant transport in groundwater
wheat growth model multi-agent system
visualization software for mathematical models
dynamic simulation of water quality urban slums in India using sensing data
tool for estimating the energy consumption
software packages for ecological systems
estimating ecological water requirement
visualization
of printer data
sustainability and interactive technologies
ecodesign tools
ecological water demand for basin systems
verify ensemble forecasts of numeric variables optimizing connectivity in conservation planning
general purpose
•
design of brush cutter
domain-specific
Figure 2.
•
knowledge discovery for coastal waters
evolution of virtual fences
Taxonomy of research topics
High domain-specifity.
Reason: The frameworks and methods we found within
the results are highly domain-specific, e.g., [12], [13].
This is also visible in the higher density of domainspecific approaches in Fig. 2.
Conclusion: Effective approaches for supporting sustainability require specific domain knowledge.
Software engineering.
Reason: There is only one approach in software engineering that explicitly addresses sustainability. It is a
reference framework with specific application in web
engineering [14].
Conclusion: An encompassing reference framework for
SE is still missing.
RQ4: How is sustainability support performed?
Constructive support for sustainability is performed
by frameworks, models, methods, and metrics (Tab. X).
Thereby, most approaches are specific to a special application domain, as visible by the density on the domain-specific
side in Fig. 2.
• Frameworks, e.g., for civil engineering [12] or contaminant transport [15]
Table X
I NCLUDED R ESULTS PER C ONTENT T YPE
Class
Constructive
Empirical
Discussion
Type of Content
Method
Model
Metrics
Framework
Tool/Prototype
Review
Study
Evaluation
Assessment
Overview
Challenges
Analysis
Reflection
Perspective
Strategy
Number of Results
36
18
1
2
9
6
8
3
4
2
1
1
1
1
1
We have derived a taxonomy for the domains that were used
and described in the publications in Fig. 4. We used the same
dimensions as for the research topics taxonomy in Fig. 2
and identified five coarse-grained domain clusters: Systems
& Knowledge in the area of general purpose, analytical
approaches, Technologies & Methods on the constructive
side of the general purpose dimension, Education somewhere in the middle between these two, special Disciplines
provide more domain-specific, analytical approaches, and
the corresponding Application & Implementation cluster
contributes the domain-specific, constructive approaches.
These clusters are not overlap-free, but only a means to
illustratively structure their diversity. The terms within the
cluster clouds in Fig. 4 indicate the individual domains.
IV. D ISCUSSION
Models, e.g., for software systems [14] or
databases [16]
• Methods for specific application areas, e.g., security
technology [13], green product design [17], or ecology
knowledge [9]
• Metrics, e.g., for sustainability in eBusiness [18]
We chose just a few of the approaches for illustration and
preferred the ones that are rather close to sustainability in
software engineering. Furthermore, there are some empirical
publications and rather few discussions.
•
RQ5: Which methods are in use?
There is a wide variety of methods in use for different
purposes — we found traditional software engineering techniques as well as domain-specific techniques and methods
from other disciplines.
Many approaches apply entity-relationship modeling,
e.g. [9], as means to represent their data, knowledge, or
information models. Neural networks are in use for dynamic
environments and simulations, e.g. [19]. Methods adapted
from other disciplines are, inter alia, cost calculations,
e.g. [20], and life cycle analysis, e.g. [21].
RQ6: Are there case studies available?
We classified publications as case studies when they were
explicitly named as such in the abstract and they were
not, for example, only containing a small illustrative case
study within a method proposal. The studies are listed
in Tab. XI. Unfortunately, none of the studies contributes
explicitly to an understanding of how to develop software
for sustainable systems, but rather to analyses of specific
application domains. Furthermore, publications that promote
studies are often method proposals illustrated in a case study
performed by the principal researcher.
RQ7: Which domains are already considered?
For an illustrated overview, please see the weighted domain cloud that visualizes the application domains in Fig. 3.
This section provides a discussion of the results and of
the threats to validity for this study.
A. Conclusions on the State of the Art
We started our search expecting to find more results to
be classified as Sustainability in Software Engineering (S in
SE in column Benefit in Tab. II-VI). As we found less than
expected for a body of knowledge on S in SE, we decided
to extend the inclusion to publications that we classified as
a research we could learn from when further investigating
sustainability in software engineering. This lead to the other
Benefit categories S Concept, S App Dom, S SW Sol, and S
HW Sol as explained in Sec. III.
In our opinion, there is still a lot of research work to
be done, especially to support the different dimensions of
sustainability from within the software engineering discipline. This can either occur in form of domain-independent
guidelines or domain-specific methods.
B. Conclusions for a Body of Knowledge
Due to these findings, our envisioned Body of Knowledge
has areas that represent the core S in SE publications, plus
areas that represent application domains with software and
hardware solutions as well as sustainability concepts from
related disciplines that we can learn from. This is illustrated
in in Fig. 5.
Body Of Knowledge
S Concept
(22 results)
Figure 5.
S SW Sol
(25 results)
S in SE
(9 results)
S App Dom
(47 results)
S HW Sol
(1 result)
Areas of the Body of Knowledge for S in SE
Table XI
C ASE S TUDIES
Author and ref.
Huang et al. [10]
Faith-Ell et al. [22]
Liu et al. [23]
Tseng et al. [24]
Yen et al. [25]
Zhang et al. [26], [27]
Jia et al. [28]
Jin et al. [29]
Domain
web engineering
cleaner production
energy
cleaner production
business research
property development
ecology
hydrology
Context
support green customers’ decision process on electronic commerce
application of environmental requirements in Swedish road maintenance
energy requirements and carbon dioxide emissions of tourism industry
evaluating a firm’s green supply chain management
management’s role in adopting green purchasing standards in industry
costs and barriers of green property development in China
urban wetland planning in Beijing
ecological water demand for basin systems
Figure 3.
Applied method
questionnaire and experiment
semi-structured interviews
index decomposition analysis
relational analysis, experiment
questionnaires
cost analysis
ecological complexity research
integrated calculation
Weighted domain cloud, created with http://www.wordle.net/
analytical (frameworks)
Systems / Knowledge
service design
life cycle analysis
digital ecosystems
technology ecosystem
ecological knowledge
Disciplines
ecosystem information management
mechatronics
business research
knowledge models
production management
ecological engineering
electronics
geography
clean production
high performance computing
Education
design research
security
e-learning
biodiversity protection
agronomy
architecture
web engineering
informatics
software engineering
nature conservation
sustainability design
environmental education
eco-culture
Application / Implementation
eBusiness
eco-informatics
data mining
mobile learning
constructive (methods)
GIS
agriculture
environmental monitoring
parallel processing
cloud
building construction
forestry
property development
Technologies / Methods
neural network
transport
urban development
decision support
energy
smart grids
groundwater management
environmental management
digital manufacturing
ecological networks
fuzzy systems
robots
lean manufacturing
green computing
energy simulation
general purpose
domain-specific
Figure 4.
Taxonomy of application domains
C. Threats to Validity
There is a number of threats to validity that we are aware
of and tried to minimize by different mitigation actions.
1) Researcher’s bias: The semi-automatic part of the
search was performed by five researchers. There could be
a researcher’s bias as the first selection was performed by
only one researcher. We minimized the effects of such a bias
by two measures:
• We explicitly stated the research questions, inclusion
and exclusion criteria, and the rationale for performing
the search.
• The first selection was reviewed and assessed by
two expert reviewers from different institutions (TUM,
UCLM). Differences were subsequently discussed, resolved and commonly agreed upon.
• The detailed analysis of the principle researcher was
reviewed by all three expert reviewers (TUM, UCLM,
UPC).
2) Search string validity: The search string validity can
be questioned under two aspects: On one hand whether it
filtered out too many publications that would have been
relevant, and on the other hand whether it included too many
irrelevant results and was, in either case, not the adequate
search string.
Indicators for too many false positives are purely hardware
papers, but as the automatically found Green IT publications
all contained part of the second parenthesis of the search
string, they were included in the results. Then, purely application in environmental domains, for example, agricultural
support systems with no explicit relation to sustainability but
relevant in case they exhibited an explicit link to sustainability in their content. Furthermore, “environment” used in the
sense of system environment, not nature — these samples
had to be excluded by hand as well as “ecosystem” used as
population of interacting systems, for example, agents.
Indicators for too many relevant exclusions were that we
found less relevant results than we would have expected.
This can either be due to a search string that was too
restrictive, to a search that was not extensive enough, or to
the fact that there is rather little published yet on that specific
topic. Not all publications we would have expected showed
up early in the search results. For example, we missed
Cabot et al. [30], as they treat goal modeling for supporting
sustainability in the context of conference organisation.
Mahaux et al. [7] were also missing in the results, with
their work on exploring sustainability requirements.4
3) Database query evaluation: We did not have any
information on which database performed which kind of
search query evaluation, and a lazy versus an eager database
query evaluation of the search string would probably have a
4 These works were not included into the results manually because we
wanted to strictly follow the SLR method. However, they will be included
in the extended version and the envisioned body of knowledge.
significant impact on the search results, considering that we
reviewed the first 100 most relevant results.
In case of a “lazy” search string evaluation, the results
might have included more references matching early parts
of the search string than compared to matching later parts.
In that case, the results might be slightly biased in terms of
favoring the terms “sustainab*” and “software engineering”
and subordinating “green” and “software systems”.
As many of the results contained the term “software system” and not “software engineering”, we are confident that
there was no bias introduced by database query evaluation.
4) Cross-validation of the search engines: We received
hardly any double entries in the automatic search results.
We would have expected some double entries in the more
general databases ScienceDirect and WebOfScience. We
decided not to use meta search engines in our first iteration
of the SLR because relying on only one meta search engine
would have made us completely dependent of the reliability
of that engine, and using various meta search engines would
have led to highly redundant results, as a pre-check showed.
Interestingly, Web of Science found Estrin [31] highly
ranked, which originates from IEEEXplore but was not
included in the IEEEXplore results (at least not within the
first 100 results). This might be a hint towards different
search query evaluation.
It would be one interesting step in future work to replicate
the searches on more databases and meta search engines and
explicitly compare the coverage.
V. C ONCLUSIONS
In this paper, we presented the results of our SLR [4] on
the research activity in sustainability in software engineering
and related topics that allow for building up a body of
knowledge. We considered 96 of 500 reviewed publications
relevant with respect to our research questions and classified
them according to content, topic, application domain, and
potential benefit for further investigation. On that basis, we
provided taxonomies for represented research topics and
application domains. As there were not as many publications explicitly presenting work on sustainability in software
engineering than expected, we propose an extended body
of knowledge for S in SE that includes related application
domains and sustainability concepts from related disciplines
that we can learn from when further investigating S in SE.
Future work is to extend the study in two directions:
on one hand by snowballing (following references) and
on the other hand via meta search engines, book search
engines, and dedicated journal searches. Probably even more
important is the challenge of making SLRs themselves “sustainable” by providing yearly updates that not only repeat
an SLR but adapt the iterations over the years according to
lessons learned from previous iterations. Thereby, we can
establish stable bodies of knowledge.
R EFERENCES
[1] B. Kitchenham and S. Charters, “Guidelines for performing
systematic literature reviews in software engineering,” tech.
rep., Software Engineering Group, Keele University, 2007.
[2] B. Penzenstadler, B. Tomlinson, and D. Richardson, “Support
environmental sustainability by requirements engineering,”
in International Workshop on Requirements Engineering for
Sustainable Systems, 2012.
[3] H. Zhang and M. Babar, “An empirical investigation of
systematic reviews in software engineering,” in Empirical
Software Engineering and Measurement (ESEM), 2011 International Symposium on, pp. 87 –96, sept. 2011.
[4] B. Penzenstadler, V. Bauer, and C. Calero, “Sustainability in
software engineering: A systematic literature review
for building up a knowledge base,” tech. rep., Technische
Universität München, 2012.
[5] United Nations World Commission on Environment and Development, “Report of the World Commission on Environment and Development: Our Common Future,” in United
Nations Conference on Environment and Development, 1987.
[6] R. Goodland, Encyclopedia of Global Environmental Change,
ch. Sustainability: Human, Social, Economic and Environmental. John Wiley and Sons, 2002.
[7] M. Mahaux, P. Heymans, and G. Saval, “Discovering Sustainability Requirements: an Experience Report,” in 17th International Working Conference on Requirements Engineering:
Foundation for Software Quality, 2011.
[8] B. Penzenstadler, V. Bauer, C. Calero, and X. Franch, “Appendix: References of the Included Results of the SLR
Sustainability in Software Engineering.” http://www4.in.tum.
de/∼penzenst/sources/ease2012 slr appendix.pdf.
[9] D. D. Pennington, I. N. Athanasiadis, S. Bowers, S. Krivov,
J. Madin, M. Schildhauer, and F. Villa, “Indirectly driven
knowledge modelling in ecology,” International Journal of
Metadata, Semantics and Ontologies , Volume 3 Issue 3, 2008.
[10] F.-H. Huang, Y.-L. Lee, and S.-L. Hwang, “E-shopping
behavior and user-web interaction for developing a useful
green website,” in Human-Computer Interaction. New Trends
(J. Jacko, ed.), vol. 5610 of Lecture Notes in Computer
Science, pp. 446–454, Springer Berlin / Heidelberg, 2009.
Springer.
[11] D. Meadows, Thinking in Systems: A Primer. Chelsea Green
Publishing, 2008.
[12] P. Kung, V. C. Chen, and A. Robinson, “Multivariate modeling for a multi-stage green building framework,” in 2011
IEEE International Symposium on Sustainable Systems and
Technology (ISSST), pp. 1–6, IEEE, May 2011.
[13] G. Zhou, G. Duan, H. Wu, and T. Yao, “Green remanufacturing engineering in structural machinery based on reverse
engineering,” in International Conference on Security Technology, 2008. SECTECH ’08, pp. 221–225, IEEE, Dec. 2008.
[14] S. Naumann, M. Dick, E. Kern, and T. Johann, “The greensoft
model: A reference model for green and sustainable software
and its engineering,” Sustainable Computing: Informatics and
Systems, no. 0, pp. –, 2011.
[15] V. Kalivarapu and E. Winer, “A multi-fidelity software framework for interactive modeling of advective and diffusive contaminant transport in groundwater,” Environmental Modelling
& Software, vol. 23, no. 12, pp. 1370 – 1383, 2008.
[16] J. B. Cushing, N. Nadkarni, M. Finch, A. Fiala, E. MurphyHill, L. Delcambre, and D. Maier, “Component-based enduser database design for ecologists,” Journal of Intelligent
Information Systems , Volume 29 Issue 1, 2007.
[17] S. Smith and C.-C. Yen, “Green product design through
product modularization using atomic theory,” Robotics and
Computer-Integrated Manufacturing, vol. 26, no. 6, pp. 790
– 798, 2010. 19th International Conference on Flexible Automation and Intelligent Manufacturing: Lean manufacturing
and Services.
[18] F. Albertao, J. Xiao, C. Tian, Y. Lu, K. Q. Zhang, and
C. Liu, “Measuring the sustainability performance of software
projects,” in 2010 IEEE 7th International Conference on eBusiness Engineering (ICEBE), pp. 369–373, IEEE, Nov.
2010.
[19] L. Du, “Expert control based on neural networks for controlling greenhouse environment,” in Computer and Computing
Technologies in Agriculture III (D. Li and C. Zhao, eds.),
vol. 317 of IFIP Advances in Information and Communication
Technology, pp. 126–132, Springer Boston, 2010. Springer.
[20] C. Ramona, “Creating and consolidating eco-economics
through financial and fiscal instruments. contribution of green
taxes,” Proceedings of the 8th WSEAS international conference on System science and simulation in engineering, 2009.
[21] M. Bravi, M. L. Parisi, E. Tiezzi, and R. Basosi, “Life cycle
assessment of a micromorph photovoltaic system,” Energy,
vol. 36, pp. 4297–4306, July 2011.
[22] C. Faith-Ell, B. Balfors, and L. Folkeson, “The application
of environmental requirements in swedish road maintenance
contracts,” Journal of Cleaner Production, vol. 14, no. 2,
pp. 163 – 171, 2006.
[23] J. Liu, T. Feng, and X. Yang, “The energy requirements
and carbon dioxide emissions of tourism industry of western
china: A case of chengdu city,” Renewable and Sustainable
Energy Reviews, vol. 15, no. 6, pp. 2887 – 2894, 2011.
[24] M.-L. Tseng and A. S. Chiu, “Evaluating firm’s green supply chain management in linguistic preferences,” Journal of
Cleaner Production, no. 0, pp. –, 2010.
[25] Y.-X. Yen and S.-Y. Yen, “Top-management’s role in adopting
green purchasing standards in high-tech industrial firms,”
Journal of Business Research, no. 0, pp. –, 2011.
[26] X. Zhang, A. Platten, and L. Shen, “Green property development practice in china: Costs and barriers,” Building and
Environment, vol. 46, no. 11, pp. 2153 – 2160, 2011.
[27] X. Zhang, L. Shen, and Y. Wu, “Green strategy for gaining competitive advantage in housing development: a china
study,” Journal of Cleaner Production, vol. 19, no. 2-3,
pp. 157 – 167, 2011.
[28] H. F. Jia, H. T. Ma, and M. J. Wei, “Urban wetland planning: A case study in the beijing central region,” Ecological
Complexity, vol. 8, pp. 213–221, June 2011.
[29] X. Jin, D. H. Yan, H. Wang, C. Zhang, Y. Tang, G. Y.
Yang, and L. H. Wang, “Study on integrated calculation of
ecological water demand for basin system,” Science Chinatechnological Sciences, vol. 54, pp. 2638–2648, Oct. 2011.
[30] J. Cabot, S. Easterbrook, J. Horkoff, L. Lessard, S. Liaskos,
and J.-N. Mazon, “Integrating sustainability in decisionmaking processes: A modelling strategy,” in Software Engineering - Companion Volume, 2009. ICSE-Companion 2009.
31st International Conference on, pp. 207 –210, May 2009.
[31] D. Estrin, “Participatory sensing: Applications and architecture,” Ieee Internet Computing, vol. 14, pp. 12–14, Jan. 2010.