摘要
利用最优的融合簇状态估计的Krein空间卡尔曼滤波方法,得到信息形式的鲁棒卡尔曼滤波。簇头节点通过所处簇的观测模型,利用信息形式的鲁棒卡尔曼滤波实现离散形式的卡尔曼滤波。簇头节点将状态估计和可逆的误差协方差矩阵传送到中心基站,中心基站融合簇状态估计产生全局状况估计。仿真结果表明,全局状态估计相对于集中状态估计(不分簇),具有更好的性能,且通信代价更低、节点寿命长。
Using the optimal integration of the cluster state estimation with the Krein space Kalman filter, we get the information form of robust Kalman filter. The cluster head node, through the observation model,realizes diserete Kalman filter by information Kalman filter. The head nodes transmit their state estimates along with the inverse of error covarionee matrix to the central base station which fuses the cluster state estimates to generate the overall global state estimate. The simulation results demonstrate that the performance of the global state estimate is comparable to the performance of the centralized state estimate. The overall global state which possess lower communication cost, long-lived node have more better performance.