Research On Algorithm Of Indoor Positioning Using Inertial Sensor And WiFi

With the change of modern living environment,people's activities in the interior are becoming more and more abundant,and more and more location-based services are applied to all aspects of life.The traditional satellite positioning system is difficult to play in the room due to the influence of building shelter.However,the current indoor positioning technology cannot meet people's requirements in terms of cost,accuracy,and implementation complexity.WiFi-based indoor positioning has attracted the attention of researchers because of its low-cost and easy-to-implement characteristics.However,due to the fact that wireless signals are susceptible to environmental interference,positioning results are unstable.The PDR(Pedestrian Dead Reckoning)uses the sensor information to estimate the trajectory of a pedestrian.The positioning accuracy is high in a short time,but the error gradually accumulates as the distance increases.In this paper,a reliable,high-precision,low-cost,easy-to-implement location system is taken as the direction,and indoor positioning technology based on WiFi and inertial sensors is studied.In the WiFi-based positioning research,an improved weighted KNN(k-Nearest Neighbor)algorithm is proposed for the shortcomings of the current WiFi fingerprint matching algorithm.The algorithm uses a weighted calculation method that is more in line with the law of wireless signal propagation to screen reference points,which can effectively reduce the influence of position ambiguity.Compared with the traditional KNN algorithm and the WKNN algorithm,the positioning accuracy increased by 24.4% and 19.4%,respectively.In the research of PDR positioning,a dynamic constraint gait detection method is designed in this paper.This method can effectively detect the pace of pedestrians by filtering out the interference caused by the jitter of the acceleration peak time and amplitude dynamic constraints.In terms of step size estimation,considering the deficiencies of the static step size estimation model,this paper improves the adaptive step size estimation model according to the pedestrian gait pattern.Compared to other step-size models,the model uses an acceleration value to participate in the calculation of the step size so that the estimated step size approaches the true step size.Considering the advantages and disadvantages of a single source of information,this paper finally proposes a combined positioning method for WiFi and sensor information fusion.The fusion method uses the extended Kalman filter to integrate the WiFi information location results with the results of the PDR track deduction,corrects the trajectory calculated by the PDR using the WiFi positioning results,and improves the accuracy of the WiFi positioning through the PDR trajectory.The comparison experiments show that the proposed fusion location method improves the problem of long-term cumulative error of PDR positioning,and its accuracy is improved by 30% compared to WiFi independent positioning.