Research On Positioning Technology Based On Time Delay Estimation In Multipath Environment

With the popular application of 5G networks and IoT(internet of things),the emergence of new devices such as UAVs and driverless vehicles has promoted the extensive research of localization technologies.Among them,wireless localization technology based on linear relationship between delay and distance in NLOS(line-of-sight)scenarios is widely used for its advantages of high accuracy,low computational complexity,and easy deployment.However,the widespread existence of multipath and NLOS errors in real-world environments poses challenges for both delay estimation and localization algorithms.In order to solve the problems that the accuracy of multipath interference and traditional time delay estimation methods is limited by sampling rate,improvements are proposed in this thesis based on the ideas of two types of time delay estimation algorithms,respectively.For the maximum likelihood delay estimation method,this thesis proposes the strategy of combining interpolated delay estimation as the initial value,and optimizes the optimal delay point search process,which effectively improves the accuracy of the algorithm and reduces the complexity.Based on the idea of subspace decomposition,this paper proposes a joint path number estimation and weighted improved MUSIC(multiple signal classification)time delay estimation algorithm.This algorithm reduces the dimensionality of the feature decomposition matrix by smoothing the correlation matrix,and uses the decomposed feature values as weights to improve the accuracy of the algorithm when the parameter estimation is inaccurate.The simulation results show that the optimized subspace delay estimation algorithm has higher resolution and effectively reduces the complexity.In order to mitigate the impact of the NLOS component on the time-delayed localization,this thesis proposes an improved localization algorithm using residual weighting.This algorithm first performs the residual calculation by using the estimated coordinates obtained from different combinations of receiving nodes.Then the inverse of the residuals is used as the weight so that the coordinate estimates with NLOS errors are given smaller weights,thus mitigating the errors caused by NLOS transmission on the localization results.Another different way is to turn the NLOS component into a measurement that helps positioning,we propose a two-stage localization based on multipath fingerprinting,which combines a classification model with a neural network.In the coarse localization phase,the region where the target is located is determined by training a random forest classification model.After the region is identified,the neural network corresponding to that region is applied to obtain more accurate localization results.A time delay fingerprint database containing NLOS and LOS paths is built in the urban model through a ray-tracing simulation tool.The simulation results within the range of show that this method is superior to the traditional NLOS positioning method and can control 90%of the positioning error within 16 m.