Feature Model Validation: A Constraint Propagation-Based Approach

Feature model validation aims to identify errors in feature models. The two major errors, called dead features and false variable features, are caused by contradictory feature relationships in a feature model. Current existing approaches use constraint satisfaction problem (CSP) and CSP solvers to identify these feature model errors. However, CSP is a NPcomplete problem and CSP solvers reveal a weak time performance. To overcome this limitation, we develop a constraint propagation based approach to identify dead features and false variable features. The correctness and efficiency of our approach is compared with a well known feature model validation tool, called FAMA, based on a number of large-size feature models which are randomly generated. The time spent for identifying feature model errors is significantly reduced from O (2 n ) required by FAMA which uses CSP solvers to O (n 2 ), while both approaches identified the same set of errors for all the evaluated feature models.