DocumentCode
258432
Title
Security Driven Requirements Refinement and Exploration of Architecture with Multiple NFR Points of View
Author
Okubo, Takanori ; Yoshioka, Nobukazu ; Kaiya, Haruhiko
Author_Institution
Fujitsu Labs. Ltd., Kawasaki, Japan
fYear
2014
fDate
9-11 Jan. 2014
Firstpage
201
Lastpage
205
Abstract
Earlier software architecture design is essential particularly when it comes to security concerns, since security risks, requirements and architectures are all closely interrelated and interacting. We have proposed the security driven twin peaks method with a mutual refinement of the requirements, and architectures. However, there are multiple alternatives to an architecture design for initial requirements, and their choices depend on non-functional requirements (NFRs), such as security, performance, and cost which often largely change. We propose a new method we call TPM-SA2 to avoid any back-track in refinement. Each architectural alternative in TPM-SA2 is refined so that it aligns with the requirements. For each refinement, the requirements can be updated vice versa. TPM-SA2 enables us to predict the impacts on the NFRs by each candidate for the architecture, and choose the most appropriate one with respect to the impact. As a result, we can define the requirements and architectures, and estimated the development costs earlier than ever.
Keywords
security of data; software architecture; TPM-SA2; initial requirements; multiple NFR points of view; nonfunctional requirements; security driven requirement exploration; security driven requirements refinement; security risks; software architecture design; Analytical models; Computer architecture; Estimation; Security; Spirals; Usability; requirement analysis; security; threat analysis; twin peaks;
fLanguage
English
Publisher
ieee
Conference_Titel
High-Assurance Systems Engineering (HASE), 2014 IEEE 15th International Symposium on
Conference_Location
Miami Beach, FL
Print_ISBN
978-1-4799-3465-2
Type
conf
DOI
10.1109/HASE.2014.35
Filename
6754606
Link To Document