Three-Dimensional Fingerprint Locating Algorithm Of RFID Based On Deep Learning

In recent years,the rapid development of IOT technology has become a hot spot.As a result,more and more attention has been paid to wireless locating technology.Indoor positioning technology has been widely studied as an important research area of IOT technology.RFID has great advantages in indoor positioning technology due to its non-line-of-sight,short delay,low cost,and large transmission range.At present,the RFID-based fingerprint localization algorithm has been widely used,but it is mainly applied to two-dimensional positioning.In the broader application of three-dimensional positioning,there are generally problems that the positioning algorithm is too costly or has low precision.In the three-dimensional positioning,fingerprint positioning method is often used.Because the essence of the fingerprint positioning method is to find the nonlinear expression of the RSSI signal and position information,in the actual application process,due to the interference of the environment,this mapping relationship may fluctuate.Therefore,under the real environment,the quantitative relationship between RSSI and distance has become a key issue in solving RFID three-dimensional fingerprint positioning technology.The deep learning model has strong deep information extraction and nonlinear modeling capabilities.In this regard,a three-dimensional RFID fingerprinting method based on deep learning is proposed in this paper.According to the characteristics of RFID fingerprinting technology,the research on using positioning network to construct a positioning system is expanded.The research in this article focuses on the following two points:(1)An RFID three-dimensional fingerprint locating method based on deep belief network is proposed.This method has high signal feature extraction capability and nonlinear mapping capability.It is applied to indoor storage positioning system and the extracted location features have a high degree of accuracy.Compared with artificial neural network method in indoor storage and positioning system,the effect of obvious advantages.(2)A complete 3-D fingerprint RFID locating model based on particle swarm optimization and deep belief network is obtained.Including the signal strength data processing based on wavelet desiccation and the RFID tag position prediction based on PSO-DBN.Optimizing the Optimal Solution of Network Parameter in Backward Propagation of DBN by Using PSO Algorithm.Through comparative experiments,it is found that the PSO-DBN based RFID positioning model is superior to the traditional neural network fingerprint positioning method in terms of prediction error,convergence speed and positioning time.