Research On Indoor Combined Positioning Algorithm Based On WIFI/PDR

In recent years,as people's demand for location services has increased,the need for indoor positioning has become more urgent.However,due to the particularity of the indoor environment,indoor positioning cannot be performed by satellite navigation systems.Based on the WIFI fingerprint location method and the Pedestrian Dead Reckoning(PDR)algorithm based on inertial sensor,it has the advantages of low cost and easy implementation.Therefore,it has received extensive attention in the field of indoor positioning and has very important research value.The WIFI-based positioning system not only provides absolute position coordinate information,but also obtains position error independent of time.However,due to the characteristics of the WIFI signal itself,the complexity of the indoor positioning environment and the interference of multipath effects,the user is in the process of positioning.There will be problems such as position jumps.However,the PDR algorithm based on inertial device obtains the indoor pedestrian position information with relatively high positioning accuracy in a short time,but there is a problem that the error accumulates with time.Based on the above two positioning algorithms,this paper deeply studies how to use the Kalman filter model to fuse WIFI and PDR information to achieve the purpose of improving indoor positioning accuracy.Firstly,the traditional K-proximity algorithm and the weighted K-proximity algorithm are applied to the WIFI fingerprint location method.However,the traditional methods have problems such as location jump and instability due to the complexity of the indoor environment and multi-path problems.Highly stable position estimation information,this paper proposes an indoor positioning algorithm based on K proximity algorithm and support vector machine algorithm.The K-proximity algorithm is used to remove the singular points of the samples in the support vector machine,and then the support vector machine algorithm is used to calculate the coordinate information of the points to be located.This paper uses the mobile phone to collect WIFI fingerprint information in the field environment,builds an offline database,and verifies the accuracy of three indoor positioning algorithm models online.The results show that the algorithm model combining K proximity algorithm and support vector machine can reduce the possibility of position jump and instability due to the interference of singular values,and can reduce the average error of position estimation to 1.701 m.Then,for the PDR-based indoor pedestrian navigation algorithm,the heading angle offset is the main cause of the error.This paper introduces an integrated heading angle estimation algorithm,which makes full use of the magnetometer,the combination of heading information provided by INS/PDR and building direction information to obtain the heading angle estimation information of each step,and provides a reliable heading for pedestrian dead reckoning information.Aiming at the complexity of pedestrian gait,an adaptive gait detection model is proposed to adapt to the asynchronous state.The model can detect the asynchronous state of pedestrians by setting the adaptive threshold and correct it by detecting the zero-speed zero-speed correction algorithm.Finally,the position based on WIFI positioning inevitably has the problem of position jump or relatively large positioning error,and the inertial device based INS/PDR algorithm will have a problem of diverging with time.In this paper,two indoor information fusion algorithm models are used respectively.The first is WIFI/INS/PDR combined positioning algorithm based on adaptive extended Kalman filter,which combines WIFI and inertial navigation device information to obtain higher precision positioning information.Taking full advantage of the nonlinearity of WIFI signal and the characteristics of pedestrian walking state,such as walking and straight walking,a WIFI/INS/PDR with adaptive unscented Kalman filter model added with noise covariance estimator is proposed.The combination positioning algorithm complements the advantages of the two to achieve the purpose of improving the indoor positioning accuracy.This paper compares the two fusion positioning algorithms with the trajectories obtained by the individual positioning and the real trajectory by collecting data in the field.The experimental results show that the WIFI/INS/PDR combined positioning algorithm based on the adaptive unscented Kalman filter model can correct the long-term error accumulation in the traditional INS/PDR algorithm and improve the instability of WIFI positioning.At the same time,due to the nonlinear characteristics of WIFI signals compared to the adaptive extended Kalman filter model,the error of the position estimate obtained value is smaller,can be reduced to have a relatively good reliability and continuity of 1.354 m.