Research On Configuration Analysis And Optimization Of Constraint Complex Feature Model In Software Product Line

Software product line engineering is a set of methods for developing similar software systems using shared core assets.Product line engineering accelerates product development by leveraging the commonalities among the product line members while delivering the benefits of lower development costs and shorter time to market.Features play a key role in the management of variability modeling.Feature model organize features,which is an abstract description of the domain.The feature selection of the feature model is to obtain a specific product that meets the user's needs,which requires a compromise between the requirement and the feature combination described in the field.Heuristic method for automatic feature selection is the commonly used feature selection method.It is often hard to obtain a valid product from a feature model configuration due to the complex constraint space.Meanwhile,the feature selection under the IMA system faces the problems of resource allocation and so on while all needs to be further solved.Aiming at the problem of the violation of constraints in using genetic algorithm to obtain product configurations from feature models,we proposed a novel crossover operator which subject to the constrains of feature models and will not introduce new violations in to offspring.We also demonstrate the idea of viewing the number of constrain violations as the main objective to be considered first.All two approaches solved the problem of less valid solution.in the final population.The novel crossover performed the cross operation subjected to all the constrains by analyzing the structure of the feature model.It led to the effective function modules exchange with respect to the meaning of crossover operation in feature selections.The environment selection plays a key role in population evolution.We discussed the constrains satisfaction in product configuration and designed the 2-D fitness criteria by viewing the number of constrain violations as the main objective to be considered first and the rest the second.By doing that,the population will evolve towards the non-constrain-violations.We also evaluated our approaches on several different scale feature models.The result show that all those two approach can significant improve the number of valid configurations in the final population.In view of the challenge of feature selection in IMA system feature model,we discussed the the modeling of resource allocation constraints and adopts multi-objective optimization algorithm to solve the problem.First,the IMA system resource allocation model is analyzed and the mapping of task system to hardware device is modeled.Next,we introduce the constraints that need to be satisfied in the resource allocation of the feature model.Then the security constraints and resource constraints were modeled.By extending the feature model to add feature attributes and distribution points,the mapping between the feature model and the resource allocation model is completed.We designed five optimization objectives to reflect the needs of multiple users,and design coding scheme and crossover mutation operator to meet the complex constraints.Finally,the validity of the method is verified by different scale models.