摘要：

针对无线传感器网络中的节点同时定位与跟踪问题,提出一种分布式融合算法。在融合底层采用基于无迹卡尔曼滤波的交互式多模型方法,克服因泰勒展开的近似误差,并在运动模式发生变化时达到理想跟踪效果。融合中心层采用内椭球逼近融合的分布式融合方法,对局部节点所获得的目标状态估计进行融合。Monte Carlo仿真结果表明,与常用的协方差交叉法相比,该算法具有更好的融合性能,对机动目标的跟踪精度提高了33.56%,并且能在跟踪目标的同时精确估计节点位置。

关键词: 节点同时定位与跟踪, 无迹卡尔曼滤波, 分布式融合, 交互式多模型, 内椭球逼近融合

Abstract:

To handle the problem of node Simultaneous Localization and Tracking(SLAT) in Wireless Sensor Network(WSN),a distributed fusion algorithm is proposed,to overcome the approximate error caused by Taylor’s expansion,and achieve the desired tracking results when the movement pattern changes.The Interactive Multiple Model(IMM) based on Unscented Kalman Filtering(UKF) is adopted as local estimator.To fuse the local estimates of target track,the Internal Ellipsoidal Approximation Fusion(IEAF) is used at the fusion center.Monte Carlo simulation results show the effectiveness of the proposed approach,which has higher fusion accuracy of 33.56% than the Covariance Intersection(CI) in the scenario of tracking a maneuvering target,while obtaining better estimates of the node location.

Key words: node Simultaneous Localization and Tracking(SLAT), Unscented Kalman Filtering(UKF), distributed fusion, interactive multiple model, internal ellipsoidal approximation fusion

中图分类号: 

• TP391