摘要
为提高传感器网络节点的定位精度,对MDS-MAP结合非线性滤波方法的多种传感器网络定位算法进行研究.根据传感器节点间距离与节点定位坐标之间存在的非线性关系,在MDS-MAP定位算法的基础上,引入扩展卡尔曼滤波(EKF)求精算法和不敏卡尔曼滤波(UKF)求精算法,对MDS-MAP求得的节点坐标进行求精.对MDS-MAP定位算法、MDS-MAP和EKF相结合的定位算法(MDS-EKF)、MDS-MAP和UKF相结合的定位算法(MDS-UKF)的定位精度进行比较.实验结果表明:EKF和UKF等非线性滤波方法的应用可以提高定位精度,在相同条件下MDS-UKF定位算法的定位精度更高并且其生成的网络拓扑图最接近于实际网络拓扑图.
New localization algorithms for wireless sensor networks which combine multidimensional scaling map （MDS-MAP） and nonlinear filtering were studied to improve the localization accuracy of sensor nodes. According to the nonlinear relationship between the sensor node distances and the node localized coordinates, the extended Kalman filter （EKF） and the unscented Kalman filter （UKF） were applied to refine the localized coordinates obtained by the MDS-MAP algorithm. The localization accuracies of these three different localization algorithms, MDS-MAP, MDS-EKF （combination of MDS-MAP and EKF） and MDS-UKF （combination of MDS-MAP and UKF）, were compared. Experimental results show that the implementation of nonlinear filtering algorithms （EKF and UKF） can improve the localization accuracy. Under the same conditions, the MDS-UKF localization algorithm achieves the best accuracy and its generated network topology is the closest to the actual network topology.
出处
《浙江大学学报（工学版）》
EI
CAS
CSCD
北大核心
2012年第5期866-872,共7页
Journal of Zhejiang University：Engineering Science
基金
国家"973"重点基础研究发展规划资助项目(2012CB720500)
国家自然科学基金资助项目(21076179)
浙江省自然科学基金项目(Y1101355)
关键词
节点自定位
MDS-MAP
扩展卡尔曼滤波
不敏卡尔曼滤波
localization of sensor nodes
multidimensional scaling-map
extended Kalman filter（EKF）
unscented Kalman filter（UKF）
