Design Of Indoor Positioning And Navigation System Based On WiFi And Sensor Fusion

With the widespread application of wireless communication technologies and intelligent terminals,the demand of mobile users for their own location information increases rapidly.At present,GPS-based outdoor positioning system has developed more mature,but can not be applied to the indoor environment owing to its complex structure,indoor positioning technology has been studied widely.At the same time,due to the advantages of wireless network technology such as fast communication and convenient deployment,as well as the built-in sensor chip in the intelligent terminal,the indoor positioning and navigation technology has been rapidly developed and extensively studied.In this thesis,indoor positioning and navigation technology based on WiFi and sensors is taken as the research object,and an improved method is proposed for the problems.An indoor positioning and navigation system based on Android platform and Apache server is realized.The details are as follows:First,aiming at the problem of lower positioning accuracy because there are a large amount of inaccurate fingerprints in the fingerprint database constructed by AP original signal intensity collected at the reference point,in the off-line measurement phase,the method of variance filter is used to establish a small noisy fingerprint database.At the same time,taking into account that the contribution of different AP's stability to positioning results is different during online positioning,the variance of AP is taken as the influencing factor in the distance formula.The traditional location fingerprinting algorithm uses Euclidean distance to measure the similarity between fingerprints,however this thesis uses Chi-square distance.Experimental results show that this algorithm can reduce the location error.Second,since estimating the position of the user for a long time by simply using the pedestrian dead reckoning algorithm will produce a larger deviation,the particle filter and the map matching method are added to the algorithm.Then by using the map information of the wall we modify the weight of each particle,and get the user's location with the weigh sum of the position of the particle swarm.When all the particles go through the wall,we use WiFi positioning to retrieve the user's position so as to activate the particle swarm.In order to effectively suppress the inherent cumulative error in the pedestrian dead reckoning algorithm,the map matching algorithm is used to set road signs at special position points.When the user goes to these road signs and matches them,the user position of the particle filter estimation is corrected.Experimental results show that this algorithm can solve the problem of cumulative error and pedestrian trajectory through the wall well.Third,in the Android platform and Apache server to achieve a based on WiFi and sensor fusion indoor positioning navigation system.The system can display the reference points and the location point directly on the indoor map.In navigation,first draw roadmap from the user current location to the target location based on the Dijkstra algorithm,then track the location of the pedestrian by using particle filter and map matching method combined with pedestrian dead reckoning algorithm.