基于神经网络的传感器网络数据汇聚模型研究

Data-Aggregation Model based on Neural Network for Wireless Sensor Network

由于定向扩散路由无线传感器网络中的数据重复传送,导致数据融合困难,制约实际应用。针对定向扩散路由传感器网络特点,提出基于定向扩散和神经网络的无线传感器网络数据汇聚模型DAM-DD&NN。借助有导师学习的BP神经网络提高感知数据的精确性、降低感知数据的时间空间冗余度;借助无导师学习的神经网络降低数据传送过程中的冗余度。理论分析和仿真结果表明,DAM-DD&NN模型能提高网络的综合性能。

The repeated data transmission in directed diffusion routing wireless sensor networks may result in difficult data fusion and this would hinder the practical. Based on the characteristics of directed-diffu- sion-routing wireless sensor network, DAM-DD＆NN model（data-aggregation model based on directed diffusion and neural network） is proposed. With the help of BP neural network with teacher leaning, the accuracy of sensing data could be improved, and the temporal redundancy and spatial redundancy of sens- ing data be reduced. And with the aid of neural network without teacher learning, the data redundancy could be reduced in the forwarding process of data packets. Theoretical analysis and simulation results show that the DAM-DD＆NN data-aggregation model could improve the comprehensive performance of wireless sensor network.