Research And Implementation Of Heterogeneous Data Acquisition And Data Convergence Of Sensor Networks

Data acquisition of The Internet of Things(IoT) needs varieties of sensor network fromdifferent vendors, the perceived packet form has not yet come to an agreement. The differentstructures and semantics of the collected data need to define the respective analytical way, whichlead to high effort, cost and time. Secondly, the layout of a large number of sensors and sustainedmassive data acquisition bring a large number of heterogeneous data. If you cannot put forward amethod of unified data convergence in the application of IoT, it will lead to slow and incompletefeedback of the information, which makes the platform of IoT difficult to achieve a multi-serviceapplication.Firstly, to resolve the problem of none united packet form of sensor data packet, in the study ofIEC61968-9(used in electricity meter) standard whose attributes and properties are extremelysimilar with sensor data, the paper proposes United Form of Sensor Data and defines UFSD-basedXML. The experiment shows that under the standard definition, data acquisition in a multi-sensornetwork has unified heterogeneous data conversion form and in multi-application system data canexchange seamlessly with high efficiency.Secondly, in order to establish stable information processing environment and unifiedinformation retrieval interface for users, this paper presents a data convergence architecture modelbased on MapReduce. The paper focuses on analysis of the various modules in the architecture. Toresolve conflict of the heterogeneous metadata information, the paper presents a heterogeneousconflict resolution method. Also the paper defines unified user query virtual database and a similarstructured query language, and then researches on view analytics and task allocation processing inthe parser module. At last, the paper researches on the MapReduce implementation, managementand connection processing in the actuator module. The result of Experiment shows that thearchitecture model can process mass data of the heterogeneous data sources and provide completeinformation to the user request in a relatively short time.