分簇算法与压缩感知下的农田信息处理

Farmland Information Processing Based on Clustering Algorithm and Compressed Sensing

为了减小无线传感器传输数据的数据量,提出了一种相关性分簇算法与压缩感知相结合的方法。首先,将节点数据显著相关性的节点划分到一个簇中;其次,在每个簇中,根据节点间的平均相关度大小来确定参考节点与非参考节点,参考点的数据结合压缩感知进行无线网络传输,而非参考点的数据可以根据与参考节点的线性回归系数计算出来;最后,对实测的温度进行仿真实验。结果表明,簇中参考节点的数据重构误差在允许范围内,对非参考节点进行线性回归计算与原始数据相比基本吻合。

In order to reduce combining with Correlation the amount of data of wireless sensor data, this paper propose a method Clustering Algorithm and compressed sensing. First, the nodes which are significantly correlated with the node data are divided into a cluster. Then, the average correlation de- gree can determine the reference node and the non reference node in each cluster, and the data of the reference node can transfer for wireless network combining with compressed sensing, while the data of the non reference node can be calculated according to the linear regression coefficient of the reference node. For simulation experiment of the measured temperature, it is proved that the data reconstruction error of the reference node in the cluster is in the allowable range, and the linear regression calcula- tion of the non reference node is basically consistent with the original data.