基于无迹卡尔曼滤波和协方差交叉融合的分层多簇无线传感器网络多速率跟踪算法

Multi-rate Tracking Algorithm for Hierarchical Multi-cluster Wireless Sensor Networks Based on Unscented Kalman Filter and Covariance Intersection Fusion

针对无线传感器网络(wireless sensor networks,WSNs)在跟踪过程中精度低、性能差等缺点,提出基于无迹卡尔曼滤波(unscented Kalman filter,UKF)和协方差交叉(covariance intersection,CI)融合的分层多簇WSNs多速率跟踪算法。将传感器分成多个簇,同一簇中的传感器可以采用不同的采样和传输速率对目标的数据进行采集和传输。采用UKF处理传感器节点采集的数据,生成局部估计,利用CI融合算法将收集到的局部估计值形成融合估计。通过设定一个附加权重因子,为真实协方差的不确定性定义一个更严格的界限。仿真验证了方法的有效性,采用多速率分层融合估计的精度更高,效果更明显。

Aiming at the shortcomings of low precision and poor performance in the tracking process of wireless sensor networks(WSNs),a hierarchical multi-cluster WSNs multi-rate tracking algorithm was proposed based on the fusion of unscented Kalman filter(UKF)and covariance intersection(CI).The sensors were divided into several clusters.Sensors in the same cluster used different sampling and transmission rates to collect and transmit the target data.First,UKF was used to process the data collected by sensor nodes and generate local estimates.Then,the CI fusion algorithm was used to form the fusion estimation from the collected local estimates.By setting an additional weight factor,a stricter limitation for the uncertainty of the real covariance was defined.The simulation results show the effectiveness of the method.The accuracy of multi-rate hierarchical fusion estimation is higher and the effect is more obvious.