因传感器网络特殊的通信方式,以及传感器节点预处理量测的时间也各有不同.常会出现源于同一目标有序的测量数据却经网络传输后无序地到达融合中心的现象,即无序量测问题。加之,现有的相关融合算法大都是在各量测数据间噪声独立情况下建立的。为此,针对一个由多个子系统组成的传感器网络无序量测系统;其中假定每个子系统均是由两个分别与融合中心同步与异步且采样率相同的传感器组成;并在考虑各传感器测量噪声相关条件下.利用顺序加权融合技术,在融合中心建立一个能实现对目标状态实时估计且在线性最小均方误差意义下最优的递推加权融合算法。理论分析与计算机仿真表明,与现有方法相比.新算法在适用范围、实时处理能力、存储量和融合估计精度等方面均有显著的优势。
Because of the special communication modes of sensor networks and the different data of pre-processing time of each sensor node, the ordered measurements from the same target often arrive in the fusion center out of sequence across sensor networks, named as "out-of-sequence" measurement (OOSM) problem. And most of fusion algorithms are presented for the independent noises assumption. This paper researches the OOSMs Sensor networks composed of multiple subsystems and each of them has two sensors where one has the sampling rate and sampling time, and another is asynchronous with the fusion center. For the case with correlated sensor and system noises, a recursive distributed weighted OOSMs fusion algorithm, which can real-timely estimate target state variables and is optimal in minimizing the trace of the error covariance matrix, is developed. The theory analysis and computer simulation both show the advantages of the proposed algorithm such as in the cases of application range, real-time processing capability, storage capability and fusion estimate aecuracv and so on.