Hierarchical distributed classification in wireless sensor networks

Wireless sensor networks promise an unprecedented opportunity to monitor the physical environment via inexpensive wireless device. Classification based on observations from distributed sensor nodes is an important application in wireless sensor networks. Developing an efficient classification method encounters many challenges in wireless sensor net works, such as accuracy and energy consumption. In this thesis, we propose a novel hierarchical distributed classification approach, in which the sensor nodes locally build the classifier and send to parent, who combines the received classifiers by generating pseudo data training data distribution is studied. The simulation results show that our approach obtains a high accuracy and saves much energy.