An Entropy-Based Approach: Handling Uncertainty in IoT Configurable Composition Reference Model (CCRM).

IoT has expanded the boundaries of the world with physical entities and virtual components as a result of the proliferation of published, invoked, and consumed IoT items. Adapting a reference model-based approach that respects the design by reuse or configuration philosophy has become a significant challenge. Hence, according to the life cycle of a connected Thing, in order to make the composition or consumption of an IoT object reusable and configurable a configurable reference composition model (CCRM) is proposed by Atlas+. Considering the complexity, scalability, heterogeneity and dynamic changes of the IoT environment, a composition model reuse will reduce costs, burdens and time spent. However, at the design time, the configurable conception mechanism applied to the composition reference model brings uncertainty related to the choice of the most relevant composition plan. This uncertainty is due to the fact that the configurable model means a restriction of the behaviour represented by an existing composition plan model. This behaviour restriction will allow only one desired composition of the reference model while eliminating unwanted ones. The uncertainty associated in selecting the optimum configuration plan from among the options is the challenge of IoT composition. In this paper, we will propose an entropy-based uncertainty measure that allows us to take into account the dynamic aspect of the model at design time and quantify this uncertainty in order to assess the predictability and efficiency of the composition plan of IoT Objects.