Design And Implementation Of Indoor Positioning And Trajectory Prediction System Based On Bluetooth Technology

With the continuous development of wireless communication technology and Internet technology,the application field of Location Based Service(LBS)has gradually extended from outdoor into indoor.Along with the diversified industrial upgrade,indoor scene becomes more and more complicated.Therefore,the integration of multiple technologies is the development trend of the indoor location service system.The main technologies that have been involved include indoor positioning,and trajectory prediction.The traditional indoor positioning technology is realized by calculating the distance between the target point and the access point which location has been known.This method is easily affected by the environment and relies heavily on hardware.Traditional trajectory prediction technology is mostly used for real-time prediction or long-term prediction in outdoor scenes.In indoor scenes,users do not move frequently,neither real-time prediction nor long-term prediction is applicable.Indoor trajectory prediction should focus on the location that users will ahead after staying for a period of time.In view of the shortcomings of existing research and the internship project in Huawei Suzhou Research Institute,the requirement of implementing a location service system in large indoor scenes will be considered,a location algorithm based on location fingerprints and feature extraction is proposed to achieve high-precision indoor location positioning.According to the mobile mode of indoor users,a trajectory prediction algorithm based on the temporal and spatial context is proposed to achieve "weak real-time" trajectory prediction,the specific research work is as follows:Firstly,the location algorithm based on location fingerprint is selected.A device deployment method that takes into account both the signal coverage and the number of devices is proposed to avoid the high dimension of the collected fingerprint data which affect the efficiency of the location algorithm.According to the final goals and constraints of the equipment deployment plan,the equipment deployment problem is converted into a combining optimization problem that cover the most user-reachable area with minimum number of equipment.Simulated annealing algorithm is selected to solve this problem.Simulation experiments are implemented and displayed.Secondly,a positioning algorithm based on feature extraction is proposed,which extracts features from the raw fingerprint data collected in the offline stage to obtain low-dimensional,highly robust training fingerprint data,and transforms the positioning problem into a classification problem in deep learning.Multi-layer artificial neural network is adapted to improve the accuracy of classification,that is,the accuracy of positioning.Fingerprint data go through feature extraction can effectively improve the accuracy of positioning verified by setting up comparative experiments.Finally,according to the analysis that "weak real-time" prediction method is more applicable to indoor scenes,a trajectory prediction algorithm based on temporal and spatial context is proposed.Trajectory data achieve from user's historical position sequence,and the spatial temporal context between trajectory points calculated by customized time and space intervals are applied to predict.The validity of "weak real-time" prediction of indoor users' movement trajectory is verified by setting up comparative experiments.Nowadays a lot of companies have invested a lot in the research of indoor location service system.This system has both application value and commercial value.It is convenient for users and can also help companies to analyze their costumer well,discover potential associations between different users,and expand the scope of service,it is of great significance to research the indoor location-based service system with the background of the rapid development of service industry.