An Expectation Maximization-aided Bayesian Beam Tracking Approach for RIS in Mobility Scenarios
Kai You Wang, Haifan Yin, Li Tan · 2024
In this paper, we address the passive beamforming problem for Reconfigurable Intelligent Surfaces (RIS)-aided communication systems in mobility scenarios. We propose an Expectation Maximization-aided Bayesian (EMB) beam tracking approach to update the reflection coefficients under time-varying channel conditions. The proposed method first leverages an approximate angular evolution model based on mixed Gaussian distribution to obtain informative prior distribution of the angles. Then the Maximum A Posteriori (MAP) criterion is exploited to predict the channel. For a uniform linear motion model in particular, we further propose a low-complexity alternative beam tracking algorithm based on historical angle information prediction. By exploiting the properties of the linear motion model and historical angle estimation, we predict the angle of departure in sparse beam space. Simulation results demonstrate that the proposed EMB beam tracking algorithm achieves approximately 5 dB gain over the state-of-the-art methods at a velocity level of 10 m/s. Furthermore, the proposed low-complexity alternative algorithm exhibits only slight performance degradation compared to the EMB algorithm.