An Evaluation of Quantitative Non-Functional Requirements Assurance Using ArchiMate

Zhengshu Zhou,Qiang Zhi,Shuichiro Yamamoto,Shuji Morisaki

doi:10.1109/access.2020.2987964

Zhengshu Zhou, Qiang Zhi + Show 2 more

Open Access

https://doi.org/10.1109/access.2020.2987964

Copy DOI

Abstract

Goal-oriented NFR (Non-Functional Requirement) assurance approaches were used to qualitatively evaluate software architectures. Assurance cases using quantitative method have not been applied to evaluate NFR assurance for software architectures. This paper presents a system architecture evaluation method which is able to conduct quantitative NFR assurance evaluation for system architecture through ArchiMate. The paper also proposes an algorithm to automate the quantitative evaluation process. A questionnaire survey among software engineers and a case study on a vehicular safety monitor system were carried out to verify the necessity of the method. Additionally, we conducted an experimental design with 18 samples divided into 2 groups with the goal of comparing how the independent variables affect the dependent variables. The results of the experiment demonstrate that the proposed method achieves better NFR evaluation effect than the traditional approach. Moreover, compared with the traditional approach, the proposed method shortens the time for NFR evaluation. The proposed method is expected to be used at the early stage of software development projects for system NFR development, such as requirements analysis, system architecture design and system modeling. At present, the method has been applied by software engineers in a practical software project.

Highlights

Software quality management is conducted throughout the whole process of software project development, including functional requirements management, and non-functional requirements management [1]
We aim to propose a method to help engineers deal with quantitative Nonfunctional requirements (NFRs) evaluation in the system architecture design phase to drive the software development process rationally
These tables show the time spent by each subject in evaluating system architecture, the number of system NFR claims made by each subject, the number of well-founded system NFR claims, rationality ratio of these NFR claims and each subject’s score in the experiment

Summary

INTRODUCTION

Software quality management is conducted throughout the whole process of software project development, including functional requirements management, and non-functional requirements management [1]. This paper will propose a new method which is possible to conduct quantitative NFR evaluation for system architecture development based on a goal decomposition framework [29], [31]. The satisfice, contribution, criticality metric and propagation rules are defined on the Softgoal Interdependency Graph (SIG) for propagating attributes from bottom layer to evaluate the safety and security assurance level of the target system architecture This approach, cannot clarify the quantitative relationships between child softgoals for goal decomposition. We aim to propose a method to help engineers deal with quantitative NFR evaluation in the system architecture design phase to drive the software development process rationally.

A PROPOSAL FOR QUANTITATIVE SYSTEM ARCHITECTURE NFR EVALUATION

CASE STUDY OF THE PROPOSED METHOD

EXPERIMENT

GROUP CONFIGURATION

Findings

DISCUSSION