Indoor Location And Motion Distance Statistics Based On UWB

Ultra-wideband technology is becoming more and more common in indoor positioning,for its indoor positioning accuracy is higher than method based on WIFI and positioning range is larger than method base on Bluetooth.The UWB positioning system based on TDOA positioning algorithm only needs to maintain clock synchronization between the base station nodes,while the clock between the mobile node and the base station does not need to be synchronized,which greatly reduces the complexity of the system design.The TDOA localization algorithm use the range difference,which calculate by measuring the signal propagation time difference between different base stations and the same mobile node,to establish a hyperbolic equation group,and then solves the equations to obtain the coordinates of the mobile node.in order to solve the problem of the equations without solutions or multiple solutions.The least square method is used in the solution process of hyperbolic equations.In a complex indoor environment,the coordinates "jitter" that occurs in a stationary state of a mobile node,that is to say,the positions calculated at different system moments in the same location are different,which brings a huge error to the motion distance statistics of the mobile node.In order to reduce the error,it is necessary to eliminate the jitter in the distance statistics process.The motion state of the mobile node of the distance statistical method based on motion state analysis is divided into four types,which are static state,unconscious motion state,short-distance round-trip motion and long-distance motion state.The data of the quiescent state does not participate in the distance statistics to preliminary remove part of the error,and then the non-stationary data is processed by means of KF and least squares method to reduce the error caused by the inaccurate division of the motion state,thereby improving the accuracy of the distance statistics rate.In order to realize the motion state division,the original coordinate sequence data is divided into multiple subsequences according to the length of time,and the trajectory length,mean square error and the number of different coordinate points of the sub-sequence are act as feature vectors for classification training,and then K-nearest neighbor,SVM and multi-layer perceptron are used for classified.In order to validity of the test method,the real data is divided to sub-sequence,which motion states are determination by classifier,and then the trajectory optimization algorithm is used to process the motionless data and calculate the distance.The experimental results show that after the motion state is divided to remove the stationary state,and used the trajectory optimization algorithm to process the data of other states.The accuracy of the calculation result of the motion distance is greatly improved.The distance statistical method based on trajectory optimization reduces the distance statistical error to 7.18%,and the distance statistical method combining the trajectory optimization method and the motion state analysis is better than the method above,and the value of the relative error falls within 1.5% and the minimum value can reach 0.9%.