无线传感器网络中基于SVR的节点数据预测算法

Node data prediction based on SVR in wireless sensor network

无线传感器网络主要用于收集环境的信息,但是由于能量的限制或者安全性等问题,存在无线传感器网络节点失效问题,一旦节点失效,将不能收集后续数据,如何预测节点将来的数据成为一个关键问题。提出一种基于支持向量回归(SVR)的节点数据预测算法,充分利用节点先前收集的数据,预测未来的数据。从仿真实验上,证明该算法的有效性和较小的预测误差率。

Wireless sensor network is mainly used in collecting the environment information, but due to limited energy and security some nodes are easy to break down, thus these nodes will not collect data. How to forecast the sequent data will become a key problem. In this paper, a data estimating algorithm based on Support Vector Regression （SVR） in Wireless Sensor Network （WSN） was proposed which used the previous data to estimate the sequent data. The simulation shows that the scheme is efficient and of less prediction error.