Data fusion method with adaptive feature weights selection based on dynamic observation geometry

In this paper, we study the selection of fusion weights in the data fusion with multi-localization systems. Traditional measurement-level fusion methods cannot adaptively adjust the fusion weights according to the observation geometry formed between the target and the measurement stations. As a result, measurement data with higher accuracy is not assigned higher fusion weights. Although the common trajectory-level fusion method can adjust the fusion weights according to the trajectory accuracy solved by a single localization system, it sacrifices the accuracy of the multi-source information fusion calculation with multi-localization systems. To solve the above problems, we propose with adaptive feature weights selection based on dynamic observation geometry. First, the measurement-level fusion and trajectory-level fusion methods are analyzed and deduced, and the Cramer-Rao lower bound is given. Second, the influencing factors of the target localization error are deduced and analyzed, and then the multi-source information fusion weight is determined. Then, a feature-level fusion method is proposed by reconstructing the Jacobian matrix. Finally, numerical simulation and field simulation experiments corroborate the effectiveness of our proposed method.