UWB Localization Algorithm Based On GRU And Beetle Antennae Search

The development of technology is very rapid today,and the emergence of more and more wireless communication technologies has made people’s requirements for positioning technology increasingly high.In the field of indoor positioning,UWB(Ultra Wide Band)technology has become a popular technology studied by scholars due to its high anti-interference ability and high-precision positioning ability.However,due to the complexity of indoor environments and the presence of factors such as multipath and Non-Line-Of-Sight(NLOS),the accuracy of localization is greatly challenged.Neural networks have powerful nonlinear mapping capabilities and can predict information results by learning data features.In the field of localization,neural network algorithms have been widely applied.The GRU(Gated Recurrent Unit)neural network can effectively learn temporal features,process variable length data,and high-dimensional features.Its application in the field of indoor positioning can improve the accuracy and stability of position estimation.In order to reduce the impact of Non-Line-Of-Sight errors on positioning results in complex environments and improve positioning accuracy,thesis proposes a fusion algorithm based on GRU network and improved Beetle Antennae Search(BAS)algorithm.This algorithm takes TDOA data values as input,trains with the true value of the target position as a label,generates a GRU network positioning model,outputs coordinate prediction values through the model,and then introduces an improved Beetle Antennae Search algorithm to accurately search for the predicted values,thereby obtaining the final positioning results.Finally,experimental simulations were conducted using Matlab to compare and analyze the proposed algorithm with traditional TDOA localization algorithms in both LOS and NLOS environments.The experimental results show that the proposed GRU-IBAS algorithm can effectively reduce Non-Line-Of-Sight errors and improve the positioning accuracy of the TDOA algorithm.The average RMSE of its positioning in NLOS environment is about 0.13 m.