This paper presents a robust near-field (NF) beam tracking algorithm for terahertz communications based on deep Q-network (DQN). Traditional NF beam tracking methods relying on mobility models are fatal in ultra-massive MIMO systems, where even the slightest error could result in beam tracking failures. Thus, the proposed algorithm aims to maintain a stable beamforming gain by tracking the mobile station through the analysis of received signals without requiring mobile dynamics. By utilizing DQN, the proposed algorithm strengthens its tracking capability from online experiences and updates the combining beam towards positions expected to maximize beamforming gain. Throughout simulations, we compare the proposed algorithm with the Bayesian filter-based NF beam tracking algorithm. The simulation results confirm the robustness of the proposed algorithm for NF beam tracking, especially for abrupt changes in mobile dynamics.

In this work, we present a deep neural network (DNN)-based indoor fingerprinting localization method with WiFi fine time measurements (FTM). The proposed method leverages the WiFi FTM and its variance as environment features to provide accurate location estimation. An $i$ -th layer DNN structure used in this paper is implemented by back propagation using an Adam optimizer. The weights and the bias of the $l-\text{th}$ layer that minimize the loss function is computed in order to minimize the positioning mean squared error (MSE). Experimental results using real-world data obtained in a typical office setting proves the efficiency of the proposed solution. The performance of the system is remarkably improved, using the $600\times 600$ hidden layer size of the DNN, we achieved an average positioning accuracy of 0.7 m and 0.9 m for the 68-th percentiles $(1-\sigma)$ and 95-th percentiles $(2-\sigma)$ respectively.