Improvement of the dependency structure of the software architecture model with risk estimation

• Published in: Journal of statistical computation and simulation 2016-03, Vol.86 (5), p.908-921
• Main Authors: Rathish Babu, T.K.S., Sankar Ram, N.
• Format: Article
• Language: English
• Subjects: Architecture (computers); Computation; Computer programs; Mathematical models; module prediction; Modules; Representations; Risk; Risk assessment; risk evaluation and prediction; Software; software architecture; Software engineering; software lifecycle; stakeholder requirements; Systems design
• ISSN: 0094-9655; 1563-5163
• DOI: 10.1080/00949655.2015.1042379

Software architecture is widely available today to define the high-level design methodology of large software systems. The growth has become a progressively significant part of the software lifecycle, due to the increasing complexity of the software being constructed. Software designers need to carefully define and reason the structural design of complex, distributed information systems, which are often encompassed of a mix of innovative and reused modules. A good, extensible and maintainable architecture often makes the difference between successful and failed projects. A good architectural representation holds the key to the efficiency of the software architecture usage and description. In this paper, an accurate software architecture model is developed using the concept of risk evaluation and prediction. In existing software architecture models, the tree-based model is not implemented. The requirements are not in a structured manner which leads to a greater computation time for execution. In the proposed method, the tree-based software architecture helps to formulate the requirements in a structured way of representation and hence it leads to a lesser computation time for execution. The module prediction strategy is applied to split up the entire requirements into sub-groups to handle the large number of stakeholder requirements. Finally, the risk assessment stage is performed to identify the risk factors of the software model and minimize the impact of the risk. The proposed software architecture results in lesser memory utilization and execution time with better performance, reliability and flexibility than the existing neuro-fuzzy performance evaluation model.