Research And Design On Sensor Network Data Management Platform

Wireless sensor network is the important part of the perception layer in the Internet of Things (IoT), laying the foundation for the emergence of IoT. The paper shows the research work of sensor data management framework based on web services technology. It also designs and implements a sensor network data management platform for sensors network and sensor data visualization, which combines with data storage system, client system and the wireless sensor network system. The framework implements a set of unified standard interface of the operation about sensor network and sensor data on the server side, in order to realize the openness and versatility of the platform.Platform server is the core of sensor network data management platform. It designs and implements the function of sensor data reception, compression and storage and then use the MapReduce distributed computing framework to complete the statistics and analysis of sensor data. Platform server stores data in the data storage system, which is a cloud storage center and made up by the Hadoop cluster and HBase distributed database. Platform server also has a visualization front end, which provides users an access of sensor network query, monitoring and management. The sensor network data management platform in this paper has following advantages and some experiments prove its performance advantages.1) Using the Hadoop cluster and HBase distributed database, it alleviates storage challenges of mass sensor data and improves performance of data reading and writing. Through the experiment, it proves that the HBase insert and query performance are higher than the MySQL database for mass sensor data.2) Improving the piecewise constant approximation algorithm by combining with the sliding window cache mechanism, it realizes the sensor data compression algorithm called SWPCA. It can reduce the time correlation of sensor data, so as to reduce data redundancy. The piecewise constant approximation algorithm uses average approximation, which eliminated the emergence of the single point error to a certain extent.3) Analyzing sensor data by MapReduce distributed computing framework. Compared with single machine computing, the cluster computing raise the efficiency of the calculated performance in mass data.4) Using social networks to share sensor data, which realizes the simple interaction between human and sensor equipment on social networks. It provides a new way of thinking to the development of urban for wisdom, wisdom campus and so on.