Model driven framework to enhance sensor network design cycle
Résumé
Recent advances in sensor network provide more and more equipment with a large amount of sensors and actuators. However, their adoptions in modern life faces several challenges. First, sensor networks design is complex and requires several steps including network sizing, hardware and software development, deployment, and data exploitation. Generally, each step is managed by a particular tool. Therefore, to
design a sensor network, it is necessary to master several tools. Furthermore, there are no links between these tools and steps, which needs making manual changes from one tool to another. These changes are costly in terms of development time and can lead to design errors. In order to address these issues, we propose a new model-driven framework to design a sensor network called Modesene. It can be used for modeling, analysis, deployment, and exploitation of sensor network. Multi-facets approach is used to integrate the multiple domains and steps in one environment. Modesene supports sensor network tools by providing bridges with these tools. It supports simulations and code generation. We validate Modesene through use case of smart building with two possible solutions. Based on Omnet++ simulator, Modesene can estimate node lifetime, costs, and packets loss. Thereafter, the obtained results drive the choice of the suitable solution based on these metrics. Then, the
code generator provides source codes for sensor nodes, gateway nodes, and also for data exploitation. Compared to existing frameworks, Modesene deals with all design steps from sizing to data exploitation.