Research On Indoor Wi-Fi Fingerprint Location Algorithm Based On Crowdsourcing

With the popularization and continuous innovation of Internet technology,the role of Wi Fi network and intelligent mobile terminal in human daily activities is becoming more and more important.At the same time,most of human activities are carried out in the indoor environment,so people's demand for location services based on indoor environment is also increasing.The satellite navigation system,which plays a leading role in outdoor positioning,has a very serious signal attenuation after penetrating the reinforced concrete wall of the building,and the accuracy also declines to 5mto20 m.Obviously,such positioning effect can not meet the indoor environment with small space,so many indoor positioning technologies emerge in time,Wi Fi fingerprint positioning technology has become a hot topic because of its low deployment cost.However,how to reduce the cost of building fingerprint database by collecting data in the positioning area and reduce the positioning offset caused by the fluctuation of Wi Fi signal itself are the focus of current research.For data collection,crowdsourcing is a good solution.By outsourcing a lot of repetitive work to other organizations or platforms,we can make full use of external resources to solve the problem.The first mock exam has been proved to reduce costs greatly in indoor positioning research.In this thesis,we first preprocess the crowdsourcing data,and propose a linear fitting method to eliminate the impact of heterogeneous devices.Then we analyze the causes of the fluctuations of Wi Fi signals,and use the moving average filter to weaken the impact as much as possible.Then we optimize the similarity model,which is very important in Wi Fi fingerprint location.At the same time,we use the improved k-means clustering to analyze the offline fingerprint Finally,the weighted k-nearest neighbor algorithm is adopted to realize the real-time measurement and the matching location of the data in the fingerprint database.In order to further improve the accuracy of Wi Fi fingerprint positioning,this thesis chooses to combine it with inertial navigation positioning.Similarly,firstly,the crowdsourced INS data is processed and converted to the navigation coordinate system.Then,the double threshold filtering algorithm is used to detect the number of steps,improve the step size model and calculate the step size.After obtaining the direction information,the extended Kalman filter is used to predict the location.Finally,a fusion strategy is proposed considering the advantages and disadvantages of Wi Fi fingerprint positioning and inertial navigation positioning At the same time,the feasibility of the fusion algorithm is verified by experiments.