Research On Indoor Fingerprint Location Algorithm Based On CSI

With the increasing popularity of wireless networks,indoor positioning technology based on WLAN has become a research hotspot due to its low equipment cost and ease of deployment.At present,most of the WLAN-based indoor positioning technologies are implemented through RSSI,but because they are susceptible to multi-path effects in indoor environments,the accuracy is not high.In this paper,the indoor positioning method based on CSI position fingerprinting is studied.The main research contents are as follows:(1)CSI data acquisition and preprocessing.First,this experiment uses Intel 5300 wireless network card and Linux 802.11n CSI-Tools open source tool to collect CSI data;After analyzing the collected CSI data,the stability information with good stability is selected as the main feature of the fingerprinting.Then,this paper uses the Pauta criterion to detect and reject abnormal data,which ensures the stability of the signal characteristics.Finally,the processed CSI data is reduced by the principal component analysis method,which reduces the amount of data computation and effectively improves the positioning efficiency.On this basis,an offline fingerprint database is established.This paper proposes a positioning algorithm based on affine propagation algorithm combined with WKNN.Firstly,according to the position fingerprint features of each reference point,the affine propagation algorithm used in this paper can cluster similar reference points into clusters without specifying the number of classes,which improves the efficiency of positioning.Then the improved WKNN algorithm is used to compare the real-time collected CSI data with the data in the offline fingerprint database.This algorithm achieves more accurate positioning by assigning different weights to each AP and its neighbors.This experiment uses MATLAB to perform simulation tests on the proposed algorithm.This article sets up an experimental scene in the laboratory and conducts several experiments.The average error of the localization algorithm proposed in this paper reaches 0.92m,which increases 39%and 52%respectively compared with the traditional KNN and Bayesian algorithm.At the same time,this paper compares the performance of CSI and RSSI in indoor positioning.Through the algorithm of this paper,the accuracy of indoor positioning based on CSI is improved by 61%compared with RSSI positioning within lm.