Research On Positioning Technology Based On RSSI And CSI

With the development of wireless communication technology and Internet of Things technology,people's demand for location-based services is increasing.Indoor location service has been applied in warehouse asset positioning,intelligent city construction,fire and earthquake disaster area emergency personnel positioning and so on.However,due to the occlusion of buildings and other reasons,GPS signals can not achieve high-precision positioning in indoor environment.Therefore,high-precision indoor positioning system and positioning scheme have become a research hotspot in the field of location services.The high popularity of WLAN makes the location technology based on WLAN easy to apply and popularize,so this location technology has been favored by researchers.At present,the wireless LAN location technology mainly includes fingerprint location technology based on RSSI(Received Signal Strength Indicator)feature and fingerprint location technology based on CSI(Channel State Information)feature.Fingerprint location technology based on RSSI features has the advantages of low feature dimension and low computational complexity of feature matching.When RSSI value is far away from the emitter,the discrimination between reference points is low,and the RSSI value fluctuates greatly at the same reference point because of the indoor multipath effect.The above two reasons lead to poor matching accuracy and low positioning accuracy of fingerprint positioning technology based on RSSI features.Compared with the deterministic method of calculating RSSI mean,the positioning method based on RSSI probability statistics has higher positioning accuracy and matching accuracy.CSI contains the amplitude and phase information of multiple subcarriers,which can represent the signal state more completely.At the same time,because there are many obstacles in the room,each reference point has different multi-path channel characteristics,and the CSI feature discrimination between reference points is high.Therefore,compared with RSSI-based fingerprint positioning technology,CSI-based fingerprint positioning technology has higher positioning accuracy.However,because of the high data dimension of CSI,the fingerprint location technology based on CSI features has the disadvantage of high computational complexity of feature matching.Aiming at the problem of low positioning accuracy using single signal,this paper studies the positioning method based on RSSI and CSI,and proposes a fusion positioning method of RSSI and CSI.The main innovations and research work of this paper are as follows:(1)To solve the problem of low positioning accuracy caused by using only CSI features,this paper proposes a fusion positioning method based on RSSI and CSI.This method combines RSSI and CSI to dynamically determine the K value of WKNN algorithm,and assists CSI positioning through RSSI feature matching results,which improves positioning accuracy.In the fusion method,aiming at the high computational complexity of CSI feature fingerprint algorithm,the fusion method is divided into two stages:the initial positioning stage based on RSSI and the precise positioning stage based on CSI.In the initial positioning stage,RSSI features are used for feature matching,high matching reference points are selected to form constrained regions,and the CSI fingerprint database used for matching in the precise positioning stage is reduced,which reduces the computational complexity.In the precise positioning stage,the feature dimension of CSI is extracted by PCA to reduce the feature dimension,and then the CSI features are used for precise matching and positioning in the constrained regions.The calculation complexity is reduced while the positioning accuracy is guaranteed.(2)In the process of fusion positioning,RSSI feature fingerprint positioning technology based on probability and statistics method uses single Gaussian model to fit the probability density function of RSSI,which leads to low fitting accuracy and low matching accuracy.In this paper,a method of constructing fingerprint database using Gaussian mixture model to fit the probability density function of RSSI is proposed,which improves the fitting accuracy of RSSI probability.It improves the accuracy of feature matching and is used in the initial positioning stage of fusion positioning and assistant positioning in precise positioning.(3)In order to solve the problem of CSI feature noise and subcarrier amplitude clustering at the same reference point in the fusion positioning process,a pre-processing method based on CSI feature is proposed in this paper.In the pre-processing method,CSI data are clustered,and a new threshold function is proposed to overcome the shortcomings of traditional wavelet threshold denoising function,which improves the effect of wavelet denoising.It is also used in the precise positioning stage of fusion positioning.In order to verify the effectiveness of the positioning method proposed in this paper,a positioning experimental platform is built in the actual indoor environment and verified by experiments.The experimental results show that in the two experimental environments,the proposed fusion location method based on RSSI and CSI filters the reference points through RSSI to form the constrained region in the initial positioning stage,and the total number of the constrained reference points decreases by 31.3%and 38.4%,respectively.Under the condition that the feature dimension of CSI is reduced by 37.8%and 36.3%,the average positioning errors are 0.93m and 1.58m,respectively.Compared with FIFS,the positioning accuracy is improved by 19.8%and 26.5%respectively,and the execution time of positioning in both scenarios is reduced by 21.6%.The experimental results show that the positioning method proposed in this paper can ensure good positioning accuracy and effectively reduce the execution time of positioning.