无线传感器网络中分布式数据挖掘算法研究

Stdudy on Distributed Data Mining Algorithm in Wireless Sensor Networks

研究无线传感器网络中数据挖掘问题。针对大量高维冗余且不确定的无线传感器网络监测数据传送到中央服务器上使用传统的数据挖掘方法进行挖掘的缺点相当明显,导致耗费了珍贵的网络能量。为解决上述问题,提出在每个传感器节点上进行基于粗糙集与遗传BP网络相结合的分布式数据挖掘算法,利用GA来训练BP网络的阈值和权值来构成遗传BP网络(GABP),克服BP网络的某些不足;利用粗糙集RS在消除冗余信息、处理不确定性数据等方面的优势,缩减训练数据输入BP网络的维度。仿真结果表明,改进算法能有效地实现对无线传感器网络中的数据进行分类,取得了较理想的效果。

This paper mainly researched data mining problems in wireless sensor network. When large high dimension redundancy and uncertain monitoring data are transfered to a central server, if using the traditional technology for data mining, it consumes valuable network energy. To solve the problem, we proposed to make distributed data mining based on rough sets and genetic BP network in every sensor node. In order to overcome some shortcoming of BP neural network, a genetic algorithm was used in the algorithm to train weight values and threshold values of BP neural network to form genetic BP neural network; and rough set theory was used to reduce the dimension of training data in BP neural network. Simulation results show that the algorithm can sort data effectively in WSN, which demonstrates the new algorithm is more efficient and can get perfect results.