Capturing Requirement Correlation in Adaptive Systems

An adaptive system is expected to modify its behavior to suit changes in environmental and system condition. Somehow, it causes problems during requirement specification (and in subsequent verification) since it is difficult to provide all possible adaptation needed during runtime. Thus, it may be necessary to temporarily ignore non-critical requirements to a certain point in order to maintain satisfaction, especially on critical and invariant requirements.One way to handle uncertainty in the adaptive system is by relaxing requirement and present the verification result using a graded (fuzzy) condition in a requirement satisfaction. Often, relaxing one requirement can affect the satisfaction of another related requirement. In this paper, we use linear regression to capture the relationship between two relaxed system requirements. Pearson and Spearman correlation coefficient is utilized to calculate correlation strength. To illustrate the approach, we consider a smart vacuum system problem.