Research On Indoor Wi-Fi Fingerprinting Localization Algorithm With Deep Learning

With the proliferation of smart devices such as wireless local area networks,smartphones,and smartwatches,indoor positioning technology has entered a stage of rapid development.Wi-Fi wireless networks have covered cities and rural areas of all sizes,and due to its own technical characteristics,many scholars have started researching Wi-Fi indoor positioning technology and proposed various location fingerprint positioning algorithms.However,traditional Wi-Fi indoor positioning suffers from drawbacks such as Wi-Fi signals being greatly affected by interference,low positioning accuracy,and susceptibility to factors such as buildings and obstruction.Therefore,there is still significant room for improvement in the performance of traditional Wi-Fi positioning algorithms.This article investigates several key technologies of deep learning-based indoor Wi-Fi fingerprint positioning algorithms.Wi-Fi fingerprint recognition typically assumes a positive correlation between real-world distance and Received Signal Strength(RSS)distance.However,due to the complexity of indoor environments,this assumption is rarely satisfied in practice,resulting in large positioning errors.To address this issue,this paper proposes a new fingerprint recognition method based on deep metric learning.This method uses a learned mapping function to convert the original RSS measurement values into features,making the selected candidate neighbors closer to the true neighbors in Euclidean space,thereby improving positioning accuracy.The specific research work of this article is as follows:Firstly,analyze the impact of using different Wi-Fi fingerprint data representation methods on positioning accuracy.Secondly,use multiple deep learning methods to extract features from Wi-Fi fingerprint data.Third,introduce and analyze indoor Wi-Fi fingerprint positioning technology.Based on traditional indoor Wi-Fi fingerprint positioning algorithms,a new Wi-Fi fingerprint indoor positioning algorithm based on deep metric learning is proposed.The design ideas and specific implementation methods of the algorithm are explained,and experiments are conducted in a real environment.Fourth,in order to verify the effectiveness and advancement of the algorithm in this paper,we compare it with common machine learning methods and the latest deep learning-based methods.The experiment was carried out in a complex indoor environment,and experimental results show that our method outperforms state-of-theart methods,and can improve the positioning result by about 15% compared to the popular KNN.