Angle of arrival estimation in dynamic indoor THz channels with Bayesian filter and reinforcement learning

Abstract

This paper presents a novel algorithm to estimate the Angle of Arrival (AoA) in a dynamic indoor Terahertz channel. In a realistic application, the user equipment is often moved by the user during the data transmission and the AoA must be estimated periodically, such that the adaptive directional antenna can be adjusted to realize a high antenna gain. The Bayesian filter is applied to exploit continuity and smoothness of the channel dynamics for the AoA estimation. Reinforcement learning is introduced to adapt the prior transition probabilities between system states, in order to fit the variation of application scenarios and personal habits. The algorithm is validated using the ray launching channel simulator and realistic human movement models.