Accident analysis of software architecture in high -reliability systems: Space Based Infrared System software problems

Mohanram, Sujatha (2017) Accident analysis of software architecture in high -reliability systems: Space Based Infrared System software problems. MSc(R) thesis, University of Glasgow.

Abstract

The accident analysis of SBIRS program is conducted by gathering information for 15 years to understand the cause of the accident. The program had series of failures, workarounds were developed incrementally to solve the incidental problems over the years. This resulted in major failure in thermal vacuum testing. The architecture was reassessed, the new architecture so adopted was the wrong architecture. This is the accident this research has analyzed. The cause of the accident is analyzed thoroughly to understand the circumstances in which such an architecture was adopted.
A System analysis of the environment was conducted to understand the accident circumstances and an accident analysis was conducted to understand the influence of the systemic failures of the wrong architectural decision which is the accident analyzed. A comparative study of accident analysis methodologies was undertaken to derive the best-suited method for accident analysis. A systemic accident analysis method STAMP, which analyses the accidents caused by the influence of the environment was considered as the best fit.
The STAMP accident analysis method was adopted to understand the accident in detail. The accident analysis was performed based on the reports gathered from GAO, DOD and other sources and was confirmed for its completeness and accuracy from GAO. STPA process was adopted to conduct accident analysis in three stages – identifying control structures, changes in control structures and dynamic process model. STAMP accident analysis was improved by adding context as an additional factor.
Accidents with context as the cause of the accident were analyzed to understand the possible solutions. The realization of the importance of context as accident cause was understood and the need to enhance the accident analysis model was realized. By adding context as part of the process that needs to be transferred to ensure successful completion was suggested. An organizational model that has been successful in assessing the accidents due to the context in the different domain was studied and was suggested to be adopted as preventive accident analysis model. Finally, the wrong architectural decision being the accident is contested and argued as the accident, as currently such decisions are not considered as an accident in the industry.
This research has identified the cause of the accident to be the context in which organizations were operating. The solution suggested is to stabilize the context in one organization and replicate the stabilized context around the organizations involved in the program. The solution identifies contextual enhancement techniques used in health and safety management to build a positive culture in the organization.
Thus this research has contributed towards analyzing the architectural failure in SBIRS program by identifying an accident analysis method that best suits the case study, applied the accident analysis to the case study to understand the cause of the accident. A recommendation of enhancing the factors in accident analysis was suggested and an accident prevention technique was recommended and a process to adopt this technique was suggested.
This research has led to two further recommendations for future work. An architectural technique which would create the framework of components to prevent future architectural accidents such as this case study will be followed up. And a process to successfully pass the context in order to prevent accidents caused by organizational context will be taken further.
This research is structured to understand the problem, analyze the problem using specific accident analysis methodology related to the domain detailing the accident, comparing different domains with the similar accident cause and finally recommending an accident prevention technique which had been successful in organizations.

• Item Type: Thesis (MSc(R))
• Qualification Level: Masters
• Keywords: Accidents, architectural accident analysis, software problems, single core, multi core, Processor, Space Based Infrared Systems, STAMP, system, based Process Analysis (STPA), accident analysis, DOD, defense acquisitions, GAO, Government Accountability Office, HSE, health and safety.
• Subjects: Q Science > QA Mathematics > QA75 Electronic computers. Computer science; Q Science > QA Mathematics > QA76 Computer software; T Technology > T Technology (General)
• Colleges/Schools: College of Science and Engineering > School of Computing Science
• Supervisor's Name: Johnson, Professor Christopher
• Date of Award: 2017
• Depositing User: Mrs Sujatha Mohanram
• Unique ID: glathesis:2017-8622
• Copyright: Copyright of this thesis is held by the author.
• Date Deposited: 21 Dec 2017 14:04
• Last Modified: 15 Jan 2018 08:27
• URI: https://theses.gla.ac.uk/id/eprint/8622