Universal gated recurrent unit-based 3D localization method for ultra-wideband systems

Abstract

In this study, a universal gated recurrent unit (u-GRU)-based location estimation (LE) method is proposed to obtain the location of an ultra-wideband (UWB) transmitter. The proposed u-GRU-LE system consists of a GRU-based classifier and nine localizers for nine channel models (CMs). The classifier first predicts the CM, and then, the proper localizer is selected according to the predicted CM to estimate the location of the transmitter. Rigorous simulations are executed with various CMs. From the results, it is verified that the proposed universal GRU-based 3D localization method generally performs well irrespective of the channel environments.

Keywords

• 3D localization
• Deep learning
• Gated recurrent unit (GRU)
• Ultra-wideband (UWB) system