基于不确定度量化加权的CKF算法

Consensus-based Kalman filtering algorithm based on weighted quantitative uncertainty

针对传感器网络中每个传感器节点的邻接节点状态估计值不确定度不同的问题,提出一种基于不确定度量化加权的一致性卡尔曼滤波算法(CKF).该算法通过考虑节点度数对于传感器网络估计精度的影响,结合节点度数提出了一种衡量邻接节点状态估计值不确定度的量化函数,并把量化值作为该邻接节点与当前节点的状态估计值偏差的融合权重引入一致性协议中,利用优化后的一致性协议对传感器节点先验估计值进行更新,可提高一致性卡尔曼滤波算法的估计精度;算法同时具有非一致性误差小和鲁棒性强等特点.最后在3种不同网络类型下,通过动态目标跟踪实验仿真验证了算法的有效性.

Aiming at the different uncertainty of adjacency nodes＇ estimate value in wireless sensor networks, a novel consensus-based Kalman filtering algorithm based on weighted quantitative uncertainty was proposed. Considering the effect of node degree on estimate accuracy, a quantization function used for quantizing the uncertainty of adjacency nodes＇ estimate value was proposed firstly in the algorithm, and then the quantitative value was used to optimize the consensus protocol as a fusion weight of deviation between the adjacency nodes＇ estimate and the current noder＇s estimate. Then the optimized consensus protocol was used for updating the node＇s prior estimate value, which could improve the estimated accuracy of consensus based Kalman filtering algorithm. Furthermore, minor inconsistency error and strong robustness could be fulfilled in the novel algorithm. Finally, the effectiveness of the novel algorithm was proved by simulations of mobile target tracking problem under three different kinds of network.