Research On Position Data Model Based On Indoor Location And Platform Design

With the development of sensor technology and location technology,location service plays a more and more important role in people's life.Indoor shopping malls,airports,stations and other long time gathered a large number of people,which makes the indoor location data has a larger scale and diggability.In the indoor area,the advance planning of the pedestrian path and the advance push of the corresponding information have great practical value.Existing indoor navigation and evacuation warnings rely only on real-time location data,which can only be optimal in time or distance,and can not be better planned in advance.In this paper,indoor location,data mining and processing techniques are used to model the location data,and the algorithm for individual location prediction is proposed.Furthermore,the characteristics of the regional population are analyzed.The data model and prediction analysis provided a new reference value for the optimization of indoor navigation path,which provides a prophetic basis for the evacuation of people and has important practical significance.This paper has completed the following aspects of the work:1.Research on the Principle of Position Data Related Technology.Firstly,the requirements of location data are introduced,including data precision,set type and size.Then the method of position data processing including position data preprocessing and management is introduced.Finally,the indoor location prediction and flow forecasting model are introduced.2.An Individual Location Prediction Algorithm Based on Spatio-Temporal Feature Association.Traditional location prediction methods are generally used for outdoor location prediction,lack of state analysis of indoor individual continuous motion,while the Markov model,the lack of correlation of individual location data analysis.In this paper,we propose an individual location prediction algorithm based on temporal and spatial feature association.The user's real-time motion features and user characteristics are fused to predict the location of the target user through a user-centered circular domain model.Experimental results show that the prediction accuracy of the proposed algorithm is 15.8%and 65.2%higher than that of the correlation algorithm and the transfer algorithm.3.A Group Location Prediction Algorithm Based on Interior Area Feature.In this paper,a regional feature model is proposed to predict the population distribution in the future by neural network training.The population density and population density are calculated by the population flow and distribution in the traditional indoor area.According to the experimental analysis,the extracted regional features have better representation and applicability,and the prediction error is 50%smaller than the general features.4.On the basis of indoor positioning data access platform,the Location data platform is designed,including data processing,data storage and forecasting analysis.At the same time,a non-relational database is designed as a high-performance cache between modules.The processing logic and the access mode between the modules of the platform are analyzed,which provides function and performance guarantee for the location data model.Based on the data processing and prediction algorithm of the paper,pedestrian location prediction in indoor area within 60s and prediction of regional population distribution within 10 minutes are achieved on the location data platform.The prediction accuracy is better than other related algorithms,which has the theory and practice value.