Research On Wi-fi Indoor Positioning Algorithm Based On Deep Learning

In recent years,with the development of science and technology,location-based services have become more and more important to people.The widespread deployment and low cost of Wi-Fi in real life has made the research of indoor positioning algorithms based on Wi-Fi a technology hot spot.However,the accuracy of most Wi-Fi-based indoor positioning technologies is seriously affected by RSSI volatility and requires regular calibration or updates.These shortcomings make it difficult for indoor positioning technology to make further breakthroughs.In view of the above problems,this paper proposes corresponding Wi-Fi indoor positioning algorithms based on deep learning from different perspectives of Wi-Fi signals.On the one hand,the time fluctuation of the Wi-Fi signal causes a large difference between the Wi-Fi signal in the test set and the Wi-Fi signal in the training set.In order to ensure that the indoor positioning algorithm still has a large time interval between the test set and the training set,To obtain high positioning accuracy,this paper proposes a robust Wi-Fi fingerprint positioning algorithm based on Stacked Denoising Autoencoder(SDAE)and Multi-Layer Perceptron(MLP).SDAE is based on its strong feature learning ability from the original Wi-Fi signal to the robust Wi-Fi feature independent of time.Based on MLP's Universal Approximation Theorem,MLP realizes the regression of Wi-Fi features and target locations.Mapping.On the other hand,considering that Wi-Fi signal strength not only changes with the distance from the target,but also changes with time,this paper also considers the spatial and temporal characteristics of Wi-Fi signals.Residual Network)and Long Short-Term Memory(LSTM)spatiotemporal location algorithms.Based on CNN's powerful feature representation capability and the ability of the residual unit to prevent network degradation,the residual network is used to learn the spatial characteristics of Wi-Fi signals at the same time slice,based on the advantages of LSTM learning long-term dependence,and LSTM to learn Wi-Fi signals in The time characteristics of different time slices,and finally use the fully connected layer to achieve regression positioning.After a large number of experiments,the algorithm proposed in this paper shows excellent positioning results,is superior to traditional indoor positioning technology,is superior to machine learning-based Wi-Fi indoor positioning algorithms,and other excellent Wi-Fi indoor positioning papers.