Research On Position-level Information Fusion Algorithm Fine-grained Performance

Position-level information fusion technology is widely used in military, civilian, and many other areas. It mainly according to the position information captured by the sensors to proceed wave gate tracking, data association, state estimation and the fusion estimation, which goal is to get the final trajectory. Now, people have proposed a lot of information fusion theories and technologies, and based on which various algorithms are derived. As for a certain function, sometimes we have a plurality of algorithms to choose from, and the characteristics and scope of each of these algorithms are not the same. According to the data source in the different environments, how to select the suitable combination of algorithms and parameters, and to achieve the optimal performance of the fusion system is the forefront of research.This article is on this background, deeply study the common used algorithms of various functional modules, fine-grained decomposite the algorithms to modular parts. The modular part, by its very nature, is a separate calculation of the algorithm. The research work is mainly focused on algorithm fine-grained analysis, including the following aspects:First, in-depth study the most important parts of the information fusion process:the state estimation algorithms and data association algorithms. And analyse various functional blocks of these algorithms, looking for the similarities and differences between the same type of algorithms, and to extract and abstract mutual function and do the modular processing.Second, based on the above modular fine-grained analysis of fusion algorithms, a sensitive metric of data association is proposed, it introduces the concept of uncertainty, and gives an overall, comprehensive evaluation of the fusion process.Third, concept of information entropy is introduced to analyse the position level information fusion algorithm’s performance and proceed parameter optimization. Using the information coverage plot to describe tracking performance, based on which, we can adjust algorithms、algorithm modules and parameters to achieve fusion performance optimization.