Research On Impact Modeling And Risk Management Of Software Requirements Change

As the source of software development, software requirements are uncertainty.Requirements change gives rise to economic loss, technical uncertainty, andmanagerial complexity, which is considered as a major source of softwaredevelopment risk. A majority of software projects overrun their budgets, get theirschedules prolonged, and even fail to meet users’ requirements. The complexity ofrequirements change determines that the management of requirements change shouldbe started in early stage of software development. Further, the ability to estimate thepotential risks is also critical for controlling the risk of software development. Thispaper focuses on the impact propagation from software requirement change tosoftware architecture, and discusses the risk due to requirement change in early stagesof software development.Firstly, the methods and hot research subjects about the risk of softwarerequirements change are reviewed. And the two basic theories including softwarerequirements change and change propagation analysis are introduced.Secondly, based on the relationship between the software requirement and thesoftware architecture, the process of software requirements change and two aspectswhich entail the risk of software requirement change including probability and impactare discussed. Then, the adaptive software architecture approach which involvesrequirements, architecture and risk is put forward to handle the interactions betweenrequirements and architecture. Further, a combined relation matrix based on QFD andDSM is used to measure the adaptability of software architecture, which can considerthe risk propagated from requirements change to architecture for two levels.Thirdly, the risk of change propagation is defined in terms of change propagationprobability and change impact. Then, a probabilistic model based on DSM isestablished to evaluate the risk of change propagation from requirements to softwarearchitecture. This model supports the prediction of potential risk of changepropagation for each component and accommodates module reorganization to avoidunnecessary redesigns, which is helpful to make a choice between differentarchitecture candidates based on requirement change. Furthermore, the proposedmodel is utilized to estimate project schedule and cost. Finally, the proposed model isverified via computational experiments, in which sensitivity analysis is applied to validate the reliability of the model.Fourthly, a multi-stage dynamic analysis model of requirements change impact isput forth, which is on the basis of system dynamic. The proposed model projects theimpact of requirements change to the software architecture, and concerns the changepropagation from requirements to architecture as a sequential decision problem. Then,introducing the partitioned multiobjective risk method (PMRM), the proposed modelextends the expected value to generate the conditional expected value factions. Thismodel accounts for the dynamics of requirements change, architecture and otherstrategy to the change, and permits analysis of the risk of current-stage decisions onfuture decision in light of multiobjectve associated with cost overrun. Monte Carlo isutilized to simulate the model. Then, the trade-off between cost and risk is helpful formanager to find the preferred strategies combination.