Research On Indoor Positioning Technology Based On CSI

With the continuous development of services basedlocation information and the wide deployment of wireless networks,the demand for location information is increasing,including search path,consumer promotion,interest point discovery and so on.Using wireless signal to obtain location information of target has become a hot topic.In the field of outdoor positioning technology,the location of GPS is an effective method.However,because of the "City Canyon" effect,the signal will be affected by various buildings in the outdoor environment,resulting in serious signal fading and multi-path effect,so that the GPS method can not work normally in the interior of the environment.Other technologies can not be popularized and applied,because of difficulties in deployment,precision,and poor ability of anti-interference.By detecting the radio signal intensity,the change of signal intensity can be detected in the movement process,so the change of signal loss can be obtained,and it can be changed to the distance parameter determined the action trajectory.However,RSS(Received Signal Strength,RSS)signals will fluctuate with time,and have a strong sensitivity to environmental noise,which will cause great deviation to the positioning results,so it can not be used in practical application.In recent years,the CSI(Channel State Information,CSI)method has been greatly improved in accuracy,but most of the methods based on CSI need to establish fingerprint database for location information,so it is difficult to realize location recognition in unknown environment,and it also has great influence on the real-time of location.The main contents of this paper are as follows:1.in indoor positioning,the non sight distance propagation path has a serious influence on the signal propagation,so it is of great significance to identify the non sight path in the location area for the location accuracy.In this paper,the two classification of the distance propagation path and the non sight distance propagation path is proposed in this paper.In the experiment,the distribution characteristics and phase of the amplitude are found.The variation of the position difference under different environmental conditions is traceable.A feature classification method based on SVM is proposed to identify occlusion conditions.Experimental results show that the proposed classification method can achieve better classification results in real scenes.The 2.target is often accompanied by different angle changes and distance changes for the reference points of the known location.In this paper,the launch end of the known position is used as the reference point.A method of tracking the trajectory in unknown environment by using CSI information is proposed,and the location area is divided into some areas.The interval of the same size can be represented as the two tuples made up of(?,?),representing the distance information and angle information with AP(Access Point,AP)respectively.Using the CSI information of the physical layer as the source of the data,the data collection can be carried out by the hand held equipment(device-based)and the non hand held equipment(device-free),and the determination of the indoor location is realized by the analysis of the two-dimensional information.3.in the process of relocating the traditional indoor location method based on the fingerprint database,the reasons for the low error and low efficiency are found,and the improved method is put forward: the method of obtaining the fingerprint information from the channel information of the subchannel using the SVD method is used to reduce the accuracy error caused by the uncertain fading of the CSI.At the same time,the fingerprint contrast process is simplified and the efficiency is improved.When establishing the fingerprint database,this paper takes account of the dynamic influence of the change of the measured point on the whole fingerprint database,and reduces the positioning error.In the analysis of different scenes,it is found that when using KNN algorithm to match the fingerprint,the most appropriate K value is not only related to the geometric features of the scene,but also related to the degree of the multi-path effect of the scene,thus further improving the accuracy of the location.