Relative entropy-based Kalman filter for seamless indoor/outdoor multi-source fusion positioning with INS/TC-OFDM/GNSS

The current single data source positioning navigation systems cannot meet the high precision and high reliability required for indoor/outdoor positioning service. In this study, based on an inertial navigation system, time-and-code division-orthogonal frequency division multiplexing ranging technology and a global navigation satellite system, a relative entropy-based Kalman multi-source fusion positioning model is developed. First, multi-source numerical observation data are filtered, and the outliers are processed in data layers to improve data source reliability and to extract stable observation data. Next, the degree of the multi-source data coupling is quantified in an information layer to analyze the multi-source information coupling degree and to develop a coupling degree factor and a Kalman fusion positioning model for multi-source heterogeneous information. Tests show that this method significantly improves system positioning, navigation stability and positioning precision.