Research On GNSS,Cellular And Wi-Fi Signal Based Localization Fusion Algorithms

With the popularization of mobile devices and the improvement of positioning systems,it is high of demand for navigation and positioning.Effective positioning in different positioning scenarios can significantly improve navigation and positioning service levels.Positioning algorithms based on GNSS,Wi-Fi,and cellular network have been all widely studied for indoor and outdoor environments.The Wi-Fi-assisted GNSS location algorithm and the localization fusion algorithm based on Wi-Fi and cellular network are proposed in this thesis.According to the numbers of satellites,wireless access points and cellular base stations available,the area at which the receiver located can be categorized as outdoor area,junction area and indoor area.In the outdoor area,the GNSS based trilateration method is used to locate the receiver.In the junction area,Wi-Fi and GNSS assist and fusion algorithms based on the number of satellites and access points received is proposed.When the number of satellites is not sufficient for positioning,Wi-Fi information is used to determine the floor where the receiver is located,and then the floor of receiver is used to assist GNSS positioning.This floor determination algorithm is divided into two steps: the coarse floor differentiation and the fine floor differentiation.The match of each fingerprint point with database is not required,thus the positioning efficiency is improved by this algorithm.When sufficient satellite and wireless access point signals can be received,group Kalman filter algorithm is used to locate receiver,and positioning accuracy can be improved.In the indoor area,the receiver may not be able to receive enough satellite signals.Generally,Wi-Fi signals are used for indoor positioning because they are widely available indoor environment.In this thesis,a modified WKNN method is proposed for Wi-Fi positioning algorithm,which uses the number of access points as the weight to obtain more accurate positioning results.However,the Wi-Fi signal coverage is limited,whereas the base station signal coverage is relatively wide.Hence,a BP neural network based fusion algorithm is presented with cellular network and Wi-Fi signals.The nonlinear relationship between the signal strength and position of the fingerprint point is established.The BP neural network is trained by considering the signal strength received as the input and the coordinates of receiver as the output.This algorithm can significantly improve the indoor positioning accuracy compared with the traditional fingerprint based positioning method.The fusion localization algorithms proposed in this thesis outperform existing algorithms,which enables its wide application.