A Data Streams Analysis Strategy Based on Hoeffding Tree with Concept Drift on Hadoop System
Xin Song, Huiyuan He, Shaokai Niu, Jing Gao · 2016
The massive sensor data streams analysis in the monitoring application of internet of things is very important, especially in the environments where supporting such kind of real time streaming data storage and management. In order to support the classification of the massive sensor data streams, in this paper, a massive sensor data streams analysis strategy is proposed based on Hoeffding tree with concept drift for event monitoring application on Hadoop system. The proposed strategy is sufficient for sensor data streams classification tasks using map-reduce platform of Hadoop system. Finally, the possibilities of the strategy are demonstrated on spatial sensing data streams processing operations in comparison with existing solutions in the cloud computing environment. The simulation results show that the strategy achieves more energy savings and also ensures few amounts of sensor data retained in memory.