Research On Software Process Of Wireless Sensor Network Based On MDA And Evolutionary Computation

Model driven architecture (MDA) is a model-based software development framework andsoftware design method. It divides system requirements and specific implementation technologies,realizes the interoperability among different systems, keeps the value of the model in thetechnological evolution process, and improves the efficiency of system development process. Theapplicaton of MDA in TinyOS based wireless sensor networks can improve the flexibility,scalability, maintainability and reusability in software development of wireless sensor network.This paper presents MDA based TinyOS software development framework for wireless sensornetworks, which models software development process in different abstract levels, gradually refinesto form specific business models through model transformation mechanisims,firstly, we builduniversal domain models of wireless sensor networks, then transform domain models to platformindependent models and platform independent models to platform specific models according to themapping rules, finally transform platform specific models to nesC models. To solve theoptimization constraints and model redundant problem in model transformation process, this papercombines MDA and evolutionary algorithm, treats the platform independent model optimizationprocess as the evolution process, uses the evolutionary operations of hierarchical sorting, selectingand iterative hybrid screening to optimize the platform independent models, chooses superiorindividual models and excludes redundant individual models according to the optimization function.Verification analysis shows that the proposed software process methods in this paper canreduce the cost of software development in TinyOS based wireless sensor network application,solve the performance optimization and model constraint problems of model transformation process,and improve the efficiency of software development in TinyOS based wireless sensor networks.