High Precision Indoor Positioning System Based On Subarea Revisability

The development and improvement of the Global Positioning System(GPS)technology,GPS has been applied more and more widely in our real life,and the demand for indoor Positioning technology is also increasing.Many systems with distinct deployment costs and positioning accuracies have been developed over the past decade for indoor positioning.The method based on received signal strength(RSS)is the most widely used.However,(1)RSS is easily interfered by multipath,shadow effect,where the interference of the wall is unavoidable;(2)In the offline stage,the establishment of fingerprint database by manually collecting data information of each location point,and it not only requires high cost,but also consumes a lot of time and is inefficient,which is not practical in the dynamic environment and the environment with large location area.Therefore,in this study,we propose an indoor positioning system based on the Deep Gaussian Process Regression model of subarea revisability.In the offline phase,we divided the whole indoor environment physics into several sub-intervals,and built corresponding partial fingerprint database for each sub-interval.In the online stage,we determine which sub-area the positioning target belongs to,and then use the Deep Gaussian Process Regression model to obtain the target location information.Indoor positioning is carried out by using the Deep Gaussian Process Regression model,which only needs to measure some reference points,and reduces the time and cost of data collection in the offline stage.The model converts the RSS values into four types of characterizing values as input data and then predicts the position coordinates using DGPR.Finally,after reinforcement learning,the position coordinates are optimized.The authors conducted several experiments on simulated and real environments at Tianjin University.The experiments examined different environments,different kernels,and positioning accuracy.Finally,the experimental comparison shows that the proposed method can save the calculation time of positioning prediction while maintaining the positioning accuracy.This paper mainly focuses on the following aspects:(1)In the offline phase,the area is divided according to the degree of interference of the indoor positioning space,and a corresponding partial fingerprint database is established for each sub-interval.(2)In the online positioning stage,the partial fingerprint database established by the offline phase matches the sub-interval of the positioning point,and then the depth-based Gaussian process regression model is used for position location.(3)Using the reinforcement learning training hyperparameters to continuously correct the positioning results to adapt to the dynamic changes of the environment.(4)We also proved by experiments that the proposed method is feasible and accurate.