null

King Saud University

College of Computer and Information Sciences

Software Engineering department

Energy-Aware Software Engineering: Knowledge Identification and Organization to Support Engineers

Abdulaziz Dhafer Alqahtani
439106253
[email protected]

Advisor(s)
Dr. Noureddine Abbadeni

Master Project proposal for the degree of
MSc in [Software Engineering]
College of Computer and Information Sciences
King Saud University

[First Semester]
[22/03/1442]
[08/11/2020]

Abstract
The research focuses on energy-aware in software engineering. It is significant to consider energy consumption in the early stages of development compared to considering the deployment. Knowledge related to energy-aware software engineering was considered and analyzed so that it is structured to help software designers and developers develop software that is energy efficient. Ontologies and metamodels have been deployed separately. Ontology is considered useful as there is no need of mentioning all the steps in the software project. Meta-models have been used in software engineering to create a greener computing Energy-aware software engineering involves coming up with software that is energy efficient from development through deployment. The study will give knowledge that is organized and structured to be used by software engineers and developers.

Keywords
Green software engineering, energy-aware

Table of Contents
LIST OF ABBREVIATIONS iv
LIST OF FIGURES vi
LIST OF TABLES vii
Chapter 1: Introduction 8
1.1 Introduction 8
1.2 Ontology vs. Metamodels 9
1.3 Problem Definition 11
1.4 Project Objectives 11
1.5 Relevance to the Department 11
1.6 Outcomes and Deliverables 12
Chapter 2: LITERATURE REVIEW 13
2.1 Background 13
2.2 Related Work. ` 14
2.3 Comparison of the papers 18
2.4 discussion 20
2.5 Plan 21
References 27

LIST OF ABBREVIATIONS

ICT -Information Communication Technology
IEEE -Institute of Electrical and Electronics Engineers
WIFI -Wireless Fidelity
WCED -World Convention on Environment and Development
DVFS – Dynamic Voltage and Frequency Scaling

LIST OF FIGURES

Figure 1: Mobile Phone energy consumption

13

LIST OF TABLES

Table 1: Comparison between the papers

14

29

Chapter 1: Introduction

1.1 Introduction
Software engineering is a field that has gained fame in recent past. The increase demand for software that can carry out the different activities has led to software engineers to focus on coming up with software that focus on functionalities and not sustainability. The software has had a negative impact on the environment causing increased pollution and consumption of more energy.
Green software engineering is essential as it will focus on developing tools that are energy efficient and conserves the environment. It is noted that engineers should focus on energy-aware software engineering. This is a form of engineering that focuses on using approaches and tools that allow energy consumption to be considered first-class software design goals. It can be indicated that energy awareness for software development is needed so that the software’s energy is checked. Recently, applications that have been developed tend to consume more energy and hence harming the environment.
Energy-aware software engineering is a form of engineering that aims to develop software to minimize energy consumption by making the right decision throughout the development life cycle. The engineers must focus on sustainability during these stages of development. To attain sustainability, the software engineer must focus on using green software engineering. Green software engineering refers to incorporating the sustainability factors that help in the maintenance of the software and adherence to environmental conservation in the process of development.
Software is known to be the major consumer of energy and controls how the hardware consumes energy. Therefore, to mitigate this challenge, it is essential for software engineers to go green in their designing and development of software. It is essential to search for the effects of software that have been designed focusing on usability and not sustainability and show the importance of green software engineering.
In this project, we identify the knowledge relate to energy-aware software engineering through considering the best practices that can that help in minimize energy consumption throughout the life cycle. Also, it organizes and structures the knowledge as ontology or meta-model so that it is available and accessible for software designers and developers.

1.2 Ontology vs. Metamodels
In the past, we can note that metamodels and ontologies have been developed separately. It is noted that metamodel deals with the conceptual definition. In contrast, an ontology deals with real-world descriptors of business entities and is known to be better said to be a domain ontology (Henderson-Sellers et al., 2014). Also, it is noted that the concept that is related to ontology has been reported in the literature and shown that different ontological design. However, few works are known to show the practical term and how to design an ontology.
It is widely accepted the importance of maintaining a software product; it is noted that few ontological designs tend to focus on applying the techniques to manage the knowledge to get it. Hence, ontology is said to be a methodological process to structuring an ontology that can be used to manage knowledge in software maintenance (Serna & Serna, 2014). Metamodeling and ontologies can be compatible and linked to benefit both communities and create a contribution that is said to be a coherent underpinning theory of software engineering.
Knowledge about the design and implementation models of the current version of a software system and the knowledge is known about the application domain of software system are vital for software engineering processes that are related to design, maintenance, management, and modification of the software system (Havlice et al., 2009). Utilization of the best knowledge can increase and lead to speed in every phase of the software system life cycle and lead to an increased quality of outcome software and operation time. The knowledge can be represented in ontologies and the possibilities of using OMG specification of ontology definition metamodel as the best background of coming up with ontologies (Havlice et al., 2009).
Formal domain ontology is considered useful. It is important as one does not need to say necessary step in every software project (Gaševic, Djuric & Devedžic, 2006).  This is so since software do deal with ideas and not the self-evident physical artefacts. Ontologies often depend on well-defined and systematic powerful AI concepts such as rule-based semantics and software engineers are said to be largely unfamiliar of these. MDA application can be noted that it can help in filling the gap of ontology development on the semantic web (Gaševic, Djuric & Devedžic, 2006).
Cloud computing is known to have capability of reducing power consumption through the use of computational resources that are noted to be virtualized. These are important as they provide an application with computational resources that are on demand. Auto-scaling is noted to be important and used as a cloud computing technique that allocate computational resources dynamically so that they can match their current loads accordingly. Model driven engineering approach can be important and can help in creating a greener computing environment that helps to reduce emissions that come from superfluous idle resources (Dougherty, White & Schmidt, 2012).

1.3 Problem Definition
The research aims to identify the knowledge related to energy-aware software engineering by considering the best practices that can minimize energy consumption throughout the lifecycle. Also, the project aims organize and structure knowledge as an ontology or meta-model so that it can be used by software designers and developers.

1.4 Project Objectives
The objective of the research is to minimize energy consumption by making appropriate choices throughout the development lifecycle. The specific objectives include:
· To identify the knowledge related to energy-aware software engineering.
· To organize the knowledge and structure it possibly as an ontology or kind of meta-model, it is available and accessible for software designers and developers.

1.5 Relevance to the Department
The study is significant in the field of ICT as it aims to come up with energy-efficient products. Green software engineering is significant as it will help to know how to develop software that is energy efficient. It will help designers develop software that carries out processes with minimal pollution, promote sustainability, and protect human health. The department will focus on sustainability and improving its production of software as it is the main focus of software engineering.

1.6 Outcomes and Deliverables
The study’s primary outcome will be knowledge related to energy-aware software engineering as it considers the best practices to minimize consumption throughout the lifecycle. The deliverables will be the knowledge that is organized and well-structured for being used for software designers and developers.

Chapter 2: LITERATURE REVIEW

2.1 Background

Green software engineering is a new standard that has been a device to improve software development. The standard focuses on coming up with energy-efficient applications that will reduce pollution in the environment. As much as it is noted to be a new phenomenon, most organizations take both environmentalism and sustainability into account while coming up with their management strategies and applications. es are finding it hard to incorporate these issues into their strategies as they cannot incorporate the strategies on the systems’ hardware part.
It is significant to consider sustainability when coming up with applications and planning for other IT infrastructures used in an organization for an extended period (Pinto & Castor, 2017). Green software engineering aims to cover substantial aspects of using this software in the environment. Firstly, it aims to reduce the number of emissions released by IT systems and infrastructure (Hindle, 2016). This will ensure that sustainability and environmentalism are achieved. Secondly, green software engineering will reduce the emissions of business and production processes through the aid of IT (Beghoura, Boubetra & Boukerram, 2017).
Green software engineering focuses on achieving sustainability in designing and developing software since they are the significant factors that cause systems (hardware) to consume energy. The working of hardware is said to be determined by the software and how it was designed. often tries to distance itself from its natural science cousins, referred to as software engineering that is known to be closer to sociology and psychology. The niche that can be identified in computer science is software energy consumption and sustainable computing in software engineering. The physical measurement of energy consumption comes with the limitation of measurement and experimentation in engineering. is supposed to focus on the conservation of energy when coming up with software that will be used in the environment. The software should be able to operate in the current environment effectively. The study found out that computer science does not embrace sustainability in development and hence not producing the best software.

2.2 Related Work. `

The study by (Eder & Gallagher (2017) notes that a great deal of energy in ICT systems tends to be wasted by software regardless of how the energy-efficient the underlying hardware. In need of avoiding these wastes, the study explains that the programmers are supposed to understand the energy consumption of programs in the development process and not to wait to measure energy after the deployment process. This understanding is noted and hindered by the larger conceptual gap between the hardware, which is said to consume the energy, to the high-level language and programming abstractions.
The method that was used aimed at describing the energy modeling and energy analysis. The modeling’s primary purpose was to attribute energy values to the programming constructs so that it can identify the level of the machine instructions, intermediate code, or the source code. Energy analysis was essential and ensured that the program’s inferring energy consumption from the program semantic and an energy model was understood (Eder & Gallagher, 2017). It noted that energy analysis and modeling techniques could be incorporated into the software engineering tools to assist the programmers in optimizing the produced software’s energy consumption.

The study noted that the increase in demand for computational power is tightly coupled with high energy consumption. In the recent past, energy consumption is reduced at the hardware level (Georgiou, Rizou & Spinellis, 2019). However, the software itself is known for providing various chances for improving energy efficiency. The study focused on investigating the area of energy-aware software development and the identification of open research challenges. The research aimed to reveal the limitations, features, and tradeoffs regarding the energy-performance of software development and provide insight into the existing approaches, tools, and techniques for programming energy-efficient applications.
The research was extracted from top-tier conferences and journals that focused on energy efficiency across the software development lifecycle phases. The results showed that parallel and approximate programming, source code analysis, efficient data structures, coding practice, and use of specific programming languages could lead to an increase in energy efficiency. This can also be boosted by the use of energy monitoring tools and benchmarks that provide insight for software developers.
The study carried out on “empirical evaluation of two best practice for energy-efficient software development,” aimed at assessing the impact that was measured in terms of energy saving of the best practice for achieving software energy efficiency that was extracted from previous work (Procaccianti, Fernández & Lago, 2016). The researcher carried out an empirical experiment that was carried out in a controlled environment where different green software practices to two software applications, namely query optimization in MySQL Server and usage of “sleep” instruction in Apache webserver.
The result showed that both practices effectively improve software energy efficiency, reducing consumption by 25%. After applying the two practices, resource usage is more energy-proportion. This means that an increase in CPU usage increases the energy consumed (Procaccianti, Fernández & Lago, 2016).
It is noted that running data centers requires more data, green computing has emerged as a way of reducing power bills of data centers that have a primary goal of making software that is more energy-efficient without having to compromise the performance (Singh, Naik & Mahinthan, 2015). The study notes that developers play a significant role in controlling the energy cost of data center software in writing code. The paper aimed to show how software developers can contribute to servers’ energy efficiency by choosing efficient APIs via the optimal choice of parameters to implement file reading, file compression, and file decomposition.
The study performed extensive measurement of those operations on a Dell power Edge 2950 machine running Linux and Windows Serves. The result showed that the energy cost of various APIs for those operations is sensitive to the energy cost, saving up to 76%.
The study carried on “understanding green software development” notes that IT energy consumption represents an increasing concern in organizations’ current operations. They focus on implementing technology in every sector of their operation. Traditionally, energy efficiency was addressed by the hardware designers (Beghoura, Boubetra & Boukerram, 2017). As the hardware grew, the influence of software behavior is noted to have grown significantly. The analysis carried out showed that the hardware consumed for power according to the tasks it was aimed to carry out. For instance, the analysis of the energy consumption on the server indicated that the energy consumption increased due to an increase in the number of tasks by 40% (Georgiou, Rizou & Spinellis, 2019).
More specifically, in computers and desktop computers, the study analyzed the different technological generations using different software usage scenarios. The result indicated that the power consumption could increase up to 10% depending on the software application used on the particular desktop (Georgiou, Rizou & Spinellis, 2019). The study also compared the power consumption of the mobile device, where it compared two generations of Android mobile phones. The study found out that the same applications between the two generations have different power consumption levels. Below is a typical example of a mobile application power consumption

Figure 1: Mobile Phone energy consumption

The study showed that although there are different figures on how devices consume energy, software over energy consumption is relevant. The realization of the impacts that software has on the environment over energy consumption is relevant and improved. This new mindset about software should improve and lead to the production of software that is energy-efficient.

2.3 Comparison of the papers

Table 1: Comparison between the papers

Paper

Problem

Method

Results

Energy-Aware Software Engineering

The study focuses on showing the dependency between power consumption profile and battery utilization.

The study used Lithium-ion batteries to demonstrate two nonlinear behavior: rate capacity effect and recovery effect.

The experiment showed that the convergence speed is sometimes achievable but does not lead to a performance improvement.

Software development lifecycle of Energy Efficiency Techniques and Tools

The study investigated the current work in energy-aware software development and identified open research challenges to show the limitation, features, and tradeoffs regarding energy-performance for software development.

The study analyzed and categorized research works extracted from top-tier conferences and journals concerning energy efficiency across software development lifecycle phases.

The current work shows that parallel and approximate programming, source code analyzers, efficient data structures, coding practices, and specific programming languages can significantly increase energy efficiency.

Understanding Green Software Development

The study focused on using a deep learning-based SEC model for energy consumption

The model was evaluated using 14 open-source projects.

The experiment showed that deep learning performs better in SEC profiling than the alternative, such as random.

Green Software Engineering: The Curse of Methodology

The study focuses on the current state of software energy consumption and where it will go.

The study uses top-tier conferences and journals.

Software is consuming much energy, but through green software engineering, it is noted that the trend will change to the consumption of less energy.

2.4 discussion
In discussion, we will talk about the strength and weaknesses of the related work. This is important as it helps to choose the best work for answering the research problem.
The research on “energy-aware software engineering” gives a study in energy-consumption in hardware and software that can be used to reduce the same. The study is rich in information on how the hardware consumer energy. The study does not give way on how to implement green technology in software engineering. Secondly, the study on the software development lifecycle of energy efficiency techniques and tools focuses on the techniques used to achieve green software. It shows the tradeoffs, limitations, and features regarding software development and gives insight into the existing software. However, the study is not researching the best solution to achieve energy-efficient software.
A study also focuses on “understanding green software development” that gives an insight into how energy consumed by software and the device’s performance is essential. It shows why it is the software that consumes more energy and not the hardware. The study has a weakness in the solutions that it gives. Lastly, the study that focuses on “green software engineering: the curse of methodology” shows the relationship between computer science and sustainability. It also shows how sustainability can be achieved through the development of software. The study is not clear as expected to give how to achieve sustainability through software.
With these related works, they will be used to extract the required knowledge that will be used to achieve good software that is friendly to the environment. Also, I will use the study to organize and structure the knowledge about green software engineering that is vital for software engineers. With this knowledge, it will achieve energy-efficient applications as a combination of knowledge will give the best practices.

2.5 Plan
In next semester, I will focus on carrying implementing the research to extract knowledge from the literature review. I will also focus on discussions to find the knowledge that will be organized as ontology or meta-model to be used by software engineers and developers. Also, I will suggest future work that is supposed to be carried out in the field of software engineering. Lastly, I will document all the findings and conclusions to be used in software engineering.

Activity

Time

Chapter Three

· Proposed System
· Overview
· Conceptual Architecture

One Week

Chapter Four

· Implementation
· Results

One week

Chapter Five

· Conclusion
· Future Work

Three days

Chapter 3: PROPOSED SYSTEM

3.1 Overview
3.2 The Green Metrics

The concept of green software is exposed to constant alterations given that technology development depicts a similar development rate. This outcome provides new opportunities for both software engineers and researchers. The aspect of sustainability is of utmost importance as it is green software given that without the former, the latter proves temporary. There is limited research on how well Software Development Life Cycle (SDLC) can be beneficially aligned with Green initiatives and sustainability (Kipp et al., 2011). That said, tremendous strides have been made given that this area of study has gained significant benefit as the efforts yield ultimate rewards. One of the most important factors are the Green Metrics. It is crucially vital to take the necessary variables under keen consideration if a software is going Green. As literatures attempt to propose Green metrics that can be potentially integrated with software engineering, developers attempt to develop these vital systems to meet the desired goals. Therefore, numerous models have been published solely basing them on the Green metrics. A system model has to adhere to certain properties, green being at the center of these attributes. The importance of factor fulfillment in software engineering cannot be overstated.
3.3 Green Soft Model

For the most part, sustainability and quality assurance go hand in hand. This makes sophisticated an aspect of utter importance and how well they blend with the requirements and the software and consumers demand. Future research requires top assessment of techniques to prevent the stipulated need from becoming a stumbling block. The Green Soft Model, a crucial conceptual model that serves as reference in the development of software products, governs the life cycle in almost every significant phase (Naumann et al., 2011). The model is known to employ both the green and sustainability metrics to ensure that software follows the right criteria in engineering. It is safe to assert the metrics dictate green software engineering and therefore the model adopted in this case bears a similar weight.

3.4
Conceptual Architecture

SDLC techniques in green software engineering proves to be development approach of vital importance. From the outset, the goals should be neatly described to ensure that the green goal remains the primary target and eventually hit. Commonly, the star model is utilized by software engineers due to its flexibility. According to Costabile et al., (2009) the SDLC stage can start from any point and end similarly. Software users determine the amount of time the developed products will be used. Optimization of the software at the hand of the users requires the developer to comprehensively understand the users’ behavior and preference. Engaging the users during SDLC is hugely rewarding when it is all said and done. Gradually, the system can be enriched in pursuance of less power consumption exposing new ways to use software. The idea can come to life once individuals utilize green software concepts and attempt to develop the systems. Mostly, theories do not translate into reality.
The star model outstandingly compliments other SDLC models (Mohankumar et al., 2015). One of these scenarios entails The Capability Maturity Model and the Energy Star Model. Therefore, employing the use of the star model for SDLC seems to be a reasonable choice given the benefits.

Figure 2: Relationship between Capability Maturity Model and SDLC Energy Star Model

3.5 Requirement

The collection of requirements is arguably the most important factor in the development process. The phase ensures that all the relevant variables and green metrics are in place. Setting up an environment is another tasks that dictates the success of developing a green software. Future problems in the development can be averted once all small to the chief specifications are laid out entirely. Given the pivotal role this stage plays and the magnitude of consequences, the sustainability of the software should be made a top-priority. Nonfunctional and functional need equal exploration. In this case, one is obligated to collect requirements according to Green and Sustainability as they are both significant. The first star caters for the stage.
3.6 Design

The design strongly relies on the information/requirements collected on the first stage. Sometimes, it is difficult to align the software requirements during designing with the consumers’ anticipation. Checking every box maybe difficult, but elaborating on the design and the potential setback eliminates intricate challenges in the near future. Green and sustainability methodologies are relevant and are to be implemented. A design that meets the sustainability goals and shows soaring satisfactory level, the second star is dealt with. Markedly, meeting the Green and Sustainability goals in the design stage the software engineer should put more emphasis on system’s performance and energy efficiency.
3.7 Development (or) coding

Although it might not be the leading factor in the SDLC, coding/development is important as it seeks to bring the design to life in the best possible way. There are wide-ranging of coding styles as it with developers’ preferences. The code and the language used is also determined by the needs of the system (Mohankumar et al., 2015). Here, Green and Sustainability considerations are the main concerns. Integrating these aspects with the code and development phase allows one to mark the third star as complete. The complete code should be clean from first line to the last. Aesthetically pleasing source code has detrimental effects for future stages. For instance, the maintenance of the green software product may require another engineer to make changes. With lines all over the place, it might be difficult to debug or improve the code.
3.8 Testing

With everything above in place, testing seeks to establish whether the stipulated goals were met. This might entail energy efficiency, speed, performance, and Green and Sustainability. It is not uncommon to sacrifice one desirable attribute for another. For instance energy efficiency for speed. In this case, energy efficiency is a pivotal target goal. The fourth star stems from the assurance of top-quality green software as the testing proves.

3.9 Implementation

Implementing the new/updated green software can be achieved through a variety of ways. An engineer may opt to release the updated versions on online platforms making it accessible for the users. Nowadays, devices receive notifications once the updates or better software are out. New software may be embedded with user manuals to guide the users throughout the preliminary stages of implementation. The last star means that the implementation phase takes each goals under serious consideration to ensure the software is what it was set out to be right from the start to the finishing line.

References
Ari, A. A. A., Damakoa, I., Titouna, C., Labraoui, N., & Gueroui, A. (2017, November). Efficient and scalable aco-based task scheduling for green cloud computing environment. In 2017 IEEE International Conference on Smart Cloud (SmartCloud) (pp. 66-71). IEEE.
Beghoura, M. A., Boubetra, A., & Boukerram, A. (2017). Green software requirements and measurement: random decision forests-based software energy consumption profiling. Requirements Engineering, 22(1), 27-40.
Costabile, M. F., Mussio, P., Provenza, L. P., & Piccinno, A. (2009, March). Supporting end users to be co-designers of their tools. In International Symposium on End User Development (pp. 70-85). Springer, Berlin, Heidelberg.
Cotes-Ruiz, I. T., Prado, R. P., García-Galán, S., Muñoz-Expósito, J. E., & Ruiz-Reyes, N. (2017). Dynamic voltage frequency scaling simulator for real workflows energy-aware management in green cloud computing. PloS one, 12(1), e0169803.
Dougherty, B., White, J., & Schmidt, D. C. (2012). Model-driven auto-scaling of green cloud computing infrastructure. Future Generation Computer Systems, 28(2), 371-378.
Eder, K., & Gallagher, J. (2017). Energy-aware software. ICT-Energy Concepts for Energy Efficiency and Sustainability, 103-127.
Gaševic, D., Djuric, D., & Devedžic, V. (2006). Model driven architecture and ontology development. Springer Science & Media.
Georgiou, S., Rizou, S., & Spinellis, D. (2019). Software Development Lifecycle for Energy Efficiency: Techniques and Tools. ACM Computing Surveys (CSUR), 52(4), 1-33.
Havlice, Z., Kunstar, J., Adamuscinova, I., & Plocica, O. (2009, January). Knowledge in software life cycle. In 2009 7th International Symposium on Applied Machine Intelligence and Informatics (pp. 153-157). IEEE.
Henderson-Sellers, B. (2011). Bridging metamodels and ontologies in software engineering. Journal of Systems and Software, 84(2), 301-313.
Henderson-Sellers, B., Gonzalez-Perez, C., Mcbride, T., & Low, G. (2014). An ontology for ISO software engineering standards: 1) Creating the infrastructure. Computer Standards & Interfaces, 36(3), 563-576.
Hindle, A. (2016, March). Green software engineering: the curse of methodology. In 2016 IEEE 23rd International Conference on Software Analysis, Evolution, and Reengineering (SANER) (Vol. 5, pp. 46-55). IEEE.
Kipp, A., Jiang, T., & Fugini, M. (2011, June). Green metrics for energy-aware IT systems. In 2011 International Conference on Complex, Intelligent, and Software Intensive Systems (pp. 241-248). Ieee.
Merino, L., Kozlova, E., Nierstrasz, O., & Weiskopf, D. (2019). VISON: An Ontology-Based Approach for Software Visualization Tool Discoverability. In 2019 Working Conference on Software Visualization (VISSOFT) (pp. 45-55). IEEE.
Mohankumar, M., & Anand, M. K. (2015). A Green IT Star Model Approach for Software Development Life Cycle. International Journal of Advanced Technology in Engineering and Science, 3(1), 548-559.
Naumann, S., Dick, M., Kern, E., & Johann, T. (2011). The greensoft model: A reference model for green and sustainable software and its engineering. Sustainable Computing: Informatics and Systems, 1(4), 294-304.
Pinto, G., & Castor, F. (2017). Energy efficiency: a new concern for application software developers. Communications of the ACM, 60(12), 68-75.
Procaccianti, G., Fernández, H., & Lago, P. (2016). Empirical evaluation of two best practices for energy-efficient software development. Journal of Systems and Software, 117, 185-198.
Radu, L. D. (2017). Green cloud computing: A literature survey. Symmetry, 9(12), 295.
Serna, E., & Serna, A. (2014). Ontology for knowledge management in software maintenance. International journal of information management, 34(5), 704-710.
Singh, J., Naik, K., & Mahinthan, V. (2015). Impact of developer choices on energy consumption of software on servers. Procedia Computer Science, 62, 385-394.