Research On Indoor Positioning Algorithm Based On Wi-Fi Fingerprin

Indoor positioning,as an important research field of the Internet of Things(Io T),plays a crucial role in various areas such as commerce,agriculture,and industry.The main challenge of indoor positioning lies in how to improve the accuracy to meet people’s growing life demands.However,current algorithms still face issues such as low positioning accuracy,weak anti-interference ability,high positioning cost,and long positioning time.This study proposes solutions to the above-mentioned problems and its main work is as follows:(1)In response to the limitations of traditional indoor Wi-Fi fingerprint positioning algorithms,which only rely on a single distance metric and do not consider the relationship between d Bm values and power,this study proposes a voting-based indoor Wi-Fi fingerprint positioning algorithm.After collecting Received Signal Strength(RSS)data,the algorithm preprocesses the RSS data and selects the common neighbors of each distance metric using a voting mechanism.It then calculates the frequency of each common neighbor and obtains the final location estimate by probability weighting.Experimental results show that the proposed algorithm achieves a positioning accuracy of 1.63 meters,which is a 10%,33%,and 58%improvement over the K Nearest Neighbors(KNN),Spearman,and Kendall tau rank correlation coefficient(KTCC)methods,respectively.Additionally,compared to the optimal positioning accuracy of 1.86 meters in the MAN2 dataset,the proposed algorithm improves the accuracy by 12%.(2)To address the problem of location algorithms being susceptible to environmental interference,this study proposes a Back Propagation Neural Network(BPNN)-based indoor fingerprint positioning algorithm to combat environmental fluctuations.This method combines BPNN with the Weight-KNN(WKNN)method to improve the Fingerprint Similarity-based Indoor Localization(FSIL)method.In the offline stage,BPNN is trained to obtain the optimal BPNN parameter settings.In the online stage,the improved FSIL algorithm selects the K nearest neighbors,and the difference between the signal strength values of the K nearest neighbors and the target user is input into the BPNN network to obtain the Euclidean distance between the K nearest neighbors and the target user.Finally,the WKNN algorithm is used to obtain the user’s final location.Simulation experiments based on the Logarithmic distance path loss(LDPL)model and Wireless Insite software,as well as test results based on the IPIN2016＿Tutorial indoor localization dataset,show that the positioning accuracy in complex indoor scenes can be improved by at least 11%.(3)For Wi-Fi fingerprint-based indoor positioning methods,it is usually necessary to build a high-density location fingerprint database to ensure high-precision positioning requirements.At the same time,when the indoor area is large or the number of reference points is large and densely distributed,the computational burden in the online stage will also increase.To solve the above two problems,this study proposes an indoor fingerprint positioning algorithm based on path loss parameter estimation and Bayesian inference.In the offline stage,location fingerprints are first collected from several reference points(RPs)in different rooms,then the collected location fingerprints are filtered,and the path loss parameters(PLPs)of different rooms are trained using the logarithmic distance path loss model.Finally,the fingerprint information of other locations in the room is predicted based on the trained parameters.In the online stage,the user’s real-time fingerprint information is first filtered,then the Bayesian inference method is used to pre-determine the room where the user is located,and then the Regional Adaptive Selection(RAS)algorithm is used to select the corresponding fingerprint database for matching.Finally,the user’s final position is obtained based on the Manhattan distance.Experimental results show that the proposed algorithm improves the positioning accuracy by at least 16% compared to existing methods on Wireless Insite simulation data and the MAN dataset,and still outperforms the KNN algorithm even when the fingerprint collection workload is reduced by half in the off-line stage.