协方差交叉在分布式传感器网络定位中的应用

The application of covariance intersection in localization of DSN

节点定位的精确性在分布式传感器网络的许多应用中都起着至关重要的作用。目前较受关注的定位方法主要包括TDOA和RSS。这两种方法是非独立的,而且定位精度易受噪声影响。如果采用传统的卡尔曼滤波方式对数据加以融合,可以降低估计误差。但因假定数据间的协方差为零,使结果并非保守可靠。本文将协方差交叉算法应用于此类数据融合问题,分别在泊松分布和均匀分布情况下,对分布式传感器网络的节点定位过程加以仿真。结果显示,协方差交叉算法更加可靠,并且提高了定位精度,非常适用于分布式传感器网络。

The accuracy of node localization is crucial for many applications of Distributed Sen,or Network （DSN）. TDOA and RSS are two location technologies that are paid more attention to nowadays. But they are not independent, and are suject to noise affection. If we apply the traditional Kalman Filter method to fuse them, the estimation error could be decreased, but the result is not trustable, for Kalman Filter assumes the cross eovariance to be zero. This paper adopts Covariance Intersection （CI） algorithm into this type of data fusion, and simulates DSN nodes under Poisson and Uniform distribution separately. The results show that CI method is more reliable and can improve the location accuracy. CI algorithm is quite suitable for DSN applications.