Research On Data Fusion Oriented Context-Awareness

With the emerging of the internet of things, the interaction between things become more and more important, it requires equipments have the ability to context-awareness like human. Context-aware computing is that the computer or device can "sense" the current context by the sensors or related technology, it will make the computer look more intelligent, and also make the interaction between the devices more convenient. Context-aware computing has gotten a great deal of attention, but most of the existing context awareness models are proposed based on specific applications, and here is no unified architecture model for context-awareness.Data fusion oriented context-awareness is very important, because the data collected by the sensor is often redundancy, uncertain, heterogeneous. Such data can not directly used in the context-awareness system. Data fusion towards context-awareness is not just to eliminate redundant data, but also need to fusion the data into some messy context pieces, at last to change the context pieces to the higher level of contextual information.The main content of the study for data fusion in context-aware computing is that how to improve the traditional data fusion methods to meet the requirements of context-aware computing.In this thesis, the data fusion methods towards context-awareness are discussed. Concepts of context, context-awareness and data fusion are reviewed. The role of data fusion technology in context-aware computing and the traditional context-aware model are analyzed, and the advantages and disadvantages of Context Toolkit model are focused on. For Context Toolkit lack of data fusion layer, this thesis presents an improved model. The model using a layered architecture to divide the data fusion into two layers: the lower layer fusion and the higher layer fusion. In the lower layer, the Fuzzy logic theory and DBN are used to map sensors'raw output data into the contextual pieces. In the higher layer, the Rule-based model is used to make the sensed data completely free from ambiguity. In order to test the effect of our research results, this thesis designed an elderly monitoring system based on Android smart phone using the proposed model and methods. The experimental results show that the proposed model and the data fusion method are effective and useful. The thesis has done beneficial research work on data fusion oriented context-awareness.