Design And Implementation Of Multi-Information Fusion Indoor Positioning System Based On WiFi And Inertial Sensor

In recent years, with the rapid development and popularization of mobile Internet technology and location-based services, the demand for indoor location service is also more urgent than ever before. Because of its own limitations, the traditional single positioning technology has been unable to meet the positioning needs in the complex indoor environment.Taking into account the positioning cost, compatibility, positioning accuracy and other factors, many researchers gradually focus their attention on WiFi positioning and Pedestrian Dead Reckoning(PDR) positioning based on inertial sensors. WiFi positioning is located by the wireless received signal strength RSSI, whose stability limits its accuracy.In practical application, RSSI is easily affected by the external environment, which ends up with great volatility, which causes huge interference to positioning accuracy. PDR positioning can be more accurate because the sensors(such as inertial sensor) are not affected by the external environment easily. But this method can produce cumulative error in long time positioning.Therefore, this paper proposed a method of multi-information fusion for indoor positioning based on WiFi and inertial sensor, according to the comparison with WiFi positioning and PDR positioning. And the proposed method was studied in the aspect of optimization and information fusion for WiFi positioning and PDR positioning respectively.In WiFi positioning, in order to reduce the response time, K-means algorithm was used to build the position fingerprint database, SVM classification positioning algorithm based on WiFi fingerprint to obtain absolute position information. In PDR positioning, for pedestrian gait detection, this paper proposed the adaptive peak detection algorithm based on dynamic threshold, and sliding window algorithm was used to calculate the dynamic threshold, to realize a more accurate walking steps in statistics; for pedestrians to walk uncertainty, WiFi positioning dynamic correction step was used; in order to reduce the interfering with heading, the heading correction algorithm was used to correct the heading of pedestrians. In fusion, in order to combine the advantages of the above two positioning techniques, an fusion algorithm based on extended kalman filter was proposed,WiFi positioning was used to reduce the accumulation error of the PDR positioning, and PDR positioning was used to reduce the fluctuation of the WiFi positioning.Finally, Android positioning client based on the Android platform and positioning server terminal based on the Windows+Apache+MySQL+PHP framework was implemented. This paper designed the related experiments, which verified and evaluated the feasibility and effectiveness of the system from the positioning accuracy, pedometer accuracy, pedestrians step length estimation accuracy and comparison of the positioning trajectory. The results showed that the proposed multi-information fusion positioning method compared with single WiFi positioning and PDR positioning has obvious advantages.