Collection And Processing Technology Of Multisensor Location Fingerprint Information

Accurate location information plays an important role in daily life,such as reverse car search in parking lots,shopping mall navigation and electronic medical guidance,etc.,all require accurate indoor location information.However,due to the complex and changeable indoor environment,the signal stability used to build the location fingerprint library is easily affected.Therefore,based on the multi-sensor signal characteristics of Wi Fi,4G cellular communication network and geomagnetism,combined with location fingerprint positioning technology,this thesis studied multi-sensor Location fingerprint information collection and processing technology.The main work and innovations of this thesis are as follows:(1)Aiming at the three problems of device heterogeneity,inaccurate location marking and unstable collected data in the existing passive crowdsourcing model,the Wi Fi signal CSI phase,Wi Fi signal CSI amplitude,4G cellular communication network are used RSS and geomagnetic information are used as location fingerprint information,and an inertial sensor assisted crowdsourcing fingerprint information collection and preprocessing technology is proposed.This solution can improve the collection efficiency of location fingerprint information and the stability of RSS data.(2)Aiming at the problem of insufficient fingerprint feature mining and low positioning accuracy in location fingerprint information processing technology,combined with deep learning,a multisensor location fingerprint information processing technology based on convolutional neural network is proposed,referred to as Processing Technology of Multi-sensor Location Fingerprint InformationConvolutional Neural Network(PTMLFI-CNN).This scheme makes full use of the characteristics of multi-sensor signals,which can further improve the stability of the location fingerprint library.(3)In a real experimental environment,the performance of the two-stage technical solution for the establishment of the multi-sensor location fingerprint library proposed above was evaluated.Experimental results show that the inertial sensor-assisted crowdsourcing fingerprint information collection and preprocessing technology is superior to the existing location fingerprint information collection schemes in terms of collection efficiency and RSS data stability.In addition,the multisensor location fingerprint information processing technology based on convolutional neural network is superior to the commonly used neural network methods in both positioning stability and positioning accuracy.