The Study And Implementation Of Indoor Positioning Based On Inertial Sensors And WiFi

With the extensive coverage of WiFi wireless access points and popularity of smart phones built with inertial sensors, indoor positioning technology based on WiFi and inertial sensors has become a hot topic in the current indoor positioning research. The WiFi is widely covered, and the location techonology based on it needs no additional hardware. Howerver, the positioning accuracy is limited. Inertial sensor positioning technology has high positioning accuracy during a short period, but as time increases, there will be accumulated error. Combining these two Positioning technology not only can improve the indoor positioning accuracy, but also reduce costs, which are the common problems existing in the current indoor positioning.The paper proposed three fusion indoor positioning algorithms based on the study about the influencing factors of Wi Fi positioning and inertial sensor positioning. One is indoor positioning algorithm based on improved inertial sensors assisted by fingerprinting. The localization algorithm relys mainly on the Wi Fi indoor positioning based on fingerprint. During the off-line training stage, a large number of RSSI datas are collected to establish the fingerprint database, while during online stage, Naive Bayes algorithm is used to estimate the user’s location. A sliding window is used to detect whether the transition occurs on the process of WiFi positioning. If case hopping appears, the location with inertial sensors gives assistance, thereby improving indoor positioning accuracy. Second, the paper raises algorithm based on integral positioning assisted intelligently by fingerprint. The algorithm uses inertial integral positioning scheme, if the Wi Fi signal strength is high, we use WiFi location based on fingerprint to correct the positioning results of inertial positioning. Simultaneously, to solve the problem of integrating inertial positioning requiring preset initial position, an intelligent retrieval method is given about how to estimate the initial position and the initial direction. Third, the paper proposes the fusion algorithm based on pedestrian dead reckoning and WiFi. In order to calculate travel position, PDR uses the numerial readout of accelerometers and magnetometers to detect gait, direction and estimate step size by self-learning. When WiFi signal strength is high, we use fingerprint to correct cumulative error caused by PDR. Conversely if the signal is weak, we use a joint position of the last time locating and pedestrian dead reckoning positioning results. Experimental results show that all the three fusion algorithms improve the indoor positioning accuracy.