LAUSR

Utilizing the millimeter-wave (mmWave) frequency is a promising solution to meet fast-growing traffic demand over wireless networks. However, mmWave communications are sensitive to physical obstructions on signal propagation. In this paper, the reconfigurable intelligent surfaces (RISs) are investigated to overcome the limitations of mmWave communications. Particularly, an RIS is deployed to reflect the mmWave signals towards vehicular users who experience direct link blockages that may occur due to static or dynamic obstacles. To this end, a risk-averse optimization problem is designed to optimize the Base Station (BS) precoding matrix and the RIS phase shifts under stochastic link blockages. A solution approach is developed in two phases: the BS precoding optimization and the RIS phase shift control phases. In the first phase, a Decomposition and Relaxation-based Precoding Optimization (DRPO) algorithm is developed to obtain the optimal precoding matrix. In the second phase, a learning-based method is introduced to dynamically adjust the direction of reflected signals under channel uncertainty. Extensive simulations are presented to validate the efficacy of the developed algorithms. The obtained results show that the developed algorithms can ensure reliable transmissions to users in non-LoS areas and improve network performance.