Research Of Indoor Positioning Technology Based On WLAN Location Fingerprint And Inertial Sensor

In recent years, with the rapid development and gradual maturity of computer technology and wireless communications technology, the demand of location based service (LBS) is becoming increasing urgent in people’s daily life. Satellite navigation and positioning system can provide all-round and high-precision real-time positioning service outdoors, however, a reliable and efficient positioning technology which can satisfy people’s demand for indoor high-precision positioning has not appeared yet.Thanks to the advantages of low cost and easy to implement, the wireless local area network (WLAN) indoor positioning technology has become a hot research topic gradually. However, the complex indoor radio propagation environment makes the WLAN signal of received signal strength indication (RSSI) value shows high variability and large complexity, which leads to a low positioning accuracy. Although dead reckoning based on smartphone’s inertial sensors can get high positioning within a short time, it will cause large cumulative error when taken as separate indoor positioning system. How to take advantage of both two positioning technology and establish a combined positioning system, aims at improving the positioning accuracy, lowing cost and reducing system complexity, is one of the main content in this work.This paper focuses on several key problems to be solved in the indoor position field, namely the positioning accuracy, positioning system complexity and floor positioning, achieves the purpose of improving the positioning precision, reducing system complexity, adopts the fusion theory, has make corresponding research on WLAN location fingerprint positioning and inertial navigation positioning technology, the major contribution of this dissertation as follows:● Against the poor positioning accuracy, a combined positioning algorithm based on Gauss Kernel Function and Kalman Filtering is proposed. The Kernel can measure the similarity of RSSI between test point and reference point, so get a higher positioning precision than the traditional WKNN algorithm. After that, using kalman filtering to filter the result from kernel function algorithm, and get user’s final location. Experiments indicate that, as a contrast to WKNN algorithm, this combined positioning algorithm can improve positioning accuracy by 20.8%.● A hybrid positioning algorithm based on Wi-Fi location fingerprint positioning technology and inertial navigation positioning system is put forward in this work. First, predict user’s possible location and area using the displacement and direction supplied by initial sensors, then calculate user’s location using the combined algorithm of kernel function and kalman filtering algorithm in this area. This hybrid positioning algorithm not only can decrease the fingerprint matching computation, but also can get a high positioning accuracy by 32.8% than pure Wi-Fi location fingerprint positioning algorithm. Specially, the probability of positioning error less than lm is 54%, this combined positioning system can meet people’s high-precision positioning requirements indoors.● The floor positioning and judgment problem is studied. A floor determination method based on Support Vector Machine (SVM) is proposed. First, pick out the Access Points (AP) which have stronger floor discrimination and then establish the training sample sets, the SVM algorithm is adopted to training and supervised learning these sets, aims at getting an accurate floor determination model finally. Using the model to predict user’s floor id, experiments show that when the number of AP deployed in floor is enough, this method can get a positioning accuracy of 93.8%, even if there is less AP deploied in the floor, it can predict user’s floor correctly over 60 percent.