The Research Of Adaptation Model Construction Technique Based On Partial Behaviour Model Fusion

With the popularity of smart devices, faced with complex environment, more and moresoftware is required to have a self-adaptive ability to respond to the changing users’ needs. In thesoftware life cycle, self-adaptive software can change its behaviour according to the changes ofthe environment and users’ needs, making the software evolving in order to achieve the ultimategoal. The self-adaptive problem of software has become a hot issue in software design.Nowadays，in the self-adaptive software design, it often uses traditional module constructiontechniques. The overall adaptation model of the software system consists of a number ofadaptation modules, which improves the reusability of adaptation modules effectively, andavoids the redevelopment and redeployment issues when the requirements change. However, dueto its adaptation logic problems, such as low reusability and high complexity, it is difficult toverify the correctness and effectiveness of the model combination. In order to meet theincremental requirements of users, this paper regards improving the reusability of adaptationmodule as a starting point, introducing the formal method of partial behaviour model to thedescription of adaptation behaviour, aimed at the combination issues of adaptation logics. Basedon the pattern composition research, this paper has conducted an in-depth research, and mainlyincludes the following aspects:(1) Formal description of adaptation behaviour.When in the model description, in addition tosupport must adaptation behaviour, it should also have to point out may adaptation behaviourto support stepwise refinement. With the help of3-valued logic model description languageKMTS, it can effectively provide the unknown and reconfigurable description information of themodels,and the models to be merged can be more comparable.(2) The method of consistent model fusion. When the models are fusing, if the grammarbetween each other is conflicting, it need a lot of human intervention to eliminate such conflicts,while the merged model is still uncertain. This paper proposes the judgement method and fusion algorithm of consistent models for supporting online fusion of adaptation models.（3）Model visualization. The correctness and validity are key issues in the model fusion. Byintroducing integration tools, it makes the fusing result visualized. At the same time,we canjudge whether it has met the design requirements with the consistent fusion method.In this paper, the consistent fusion theory is applied in a model instance of the smart home.According to the fusion process and the fusion results,we analyse and verify the correctness andvalidity of the model fusion. The instance shows that building a unified adaptation behaviour canreduce the complexity of model fusion and improve the reliability of fusion results.