| With the rapid development of information technology,location based services have been widely used in shopping malls,shopping,logistics and emergency rescue.Location-based services require high-precision positioning technology to support.In the outdoor,satellite navigation and positioning technology has high-precision positioning and navigation capabilities.However,satellite signals are occluded indoors by building walls,resulting in satellite navigation and positioning technology that cannot be met.The demand for high-precision positioning in the room has made the indoor high-precision positioning technology a research hotspot.In the common indoor positioning technology,Pedestrian Dead Reckoning has the advantages of being less susceptible to external environment interference,continuous navigation and positioning,high positioning accuracy.So it become an important resear-ch direction.In order to improve the accuracy of PDR,the traditional PDR limits the attitude and position of the positioning terminal.However,the attitude and position of the positioning terminal in the actual scene are often uncertain.Therefore,PDR of the positioning terminal in multi-pose become a research problem.As time increases,there is also a cumulative error problem with PDR.Aiming at these two problems,this paper studies the PDR of the positioning terminal in multi-pose,and proposes a Multiple Attitudes based Pedestrian Dead Reckoning algorithm.At the same time,in order to eliminate the cumulative error in the PDR positioning process,a DOA/PDR fusion indoor positioning algorithm is proposed in the context of 5G Massive MIMO technology.The main resear-ch contents and innovations of this paper are as follows:Aiming at the problem that the Pedestrian Dead Reckonins is inaccurate when the positioning terminal is in multiple postures,the MAPDR algorithm is proposed.The algorithm consists of three parts:step detection,step estimation and heading angle estimation.In this paper,matrix acceleration is used to obtain the acceleration of Z-axis in geographic coordinate system.And proposed a multi-threshold constraint step detection method,which improves the accuracy of step frequency detection.The acceleration data input by the nonlinear step size model is preprocessed,and the step size estimation method of peak and valley mean filtering is proposed,which increases the accuracy of the step estimation.The frequency domain and time domain processing of X-axis and Y-axis acceleration after rotation transformation are proposed based on time-frequency transform and angle search.The pedestrian heading estimation method,and then accurately estimates the heading angle of the positioning terminal in the multi-pose descending.Aiming at the problem of accumulated error in PDR positioning process,this paper proposes a DOA/PDR fusion localization algorithm based on Kalman filter.The fusion positioning algorithm uses an accelerometer and a barometer to construct a finite state machine for judging the height of the positioning terminal,thereby implementing a single base station positioning method based on two-dimensional DOA.The positioning result of this method is combined with the extended Kalman filter of the MAPDR algorithm,which can eliminate the cumulative error of PDR.The MATLAB simulation verification platform is built.The simulated Massive MIMO base station is composed of 16×16 array antennas with a coverage of 13.2 meters.The DOA/PDR fusion localization algorithm proposed for Massive MIMO is simulated and verified.The simulation results show that the fusion positioning algorithm eliminates The cumulative error of PDR,the average positioning error is 0.16 meters,and the positioning accuracy is improved by 85.3%compared with the PDR positioning alone. |
PDF Full Text Request