| Indoor positioning is the technical foundation of location based service,which has a significant impact on human life.How to conduct pedestrian indoor positioning with real-time,cheap,flexible,convenient and high accuracy has become a research difficulty and hotspot.This paper firstly introduces the purpose and significance of the pedestrian indoor location system based on WiFi and step information.At the same time,we give an overview of the research status at home and abroad from three aspects: positioning technology,positioning algorithm and positioning model.With the tradeoff between positioning cost and positioning accuracy,this paper selects WiFi and motion sensors as the subject of indoor positioning.Since WiFi signal intensity distribution and real physical space is not continuous corresponding,this paper proposes a fusion algorithm based on the improved K-means algorithm and the improved random forest algorithm(WCluCl Model).According to the distribution characteristics of WiFi signal strength,WCluCl model uses a clustering algorithm to conduct regional partition of reference points,then in all the sub regions meeting the conditions,WCluCl model uses a classification algorithm for fine-grained location.Since WiFi fingerprint positioning ignores the pedestrian step timing and trajectory information,this paper proposes a multi-sensor fusion algorithm based on the Gaussian model,the improved KNN algorithm and a probabilistic algorithm(WaSpro model).WaSpro model obtains many candidate positions by using WiFi fingerprint,then it selects the position with the highest score as the pedestrian's current position.This paper finally develops a positioning system prototype in the Android mobile phone platform.To its function,this prototype system can be divided into three modules: functional support layer,functional implementation layer and UI layer.It is proved that the fusion positioning system is effective and feasible by experiments. |
PDF Full Text Request