Abstract
Terahertz (THz) systems can accommodate extremely high data rates thanks to large bandwidth. However, the sensitivity to blockages is a key challenge for THz signals, and even user position variation may cause misalignment of THz pencil-like beamforming resulting in sharp decline in system performance. In this paper, we propose a novel system performance optimization scheme for environment-aware Simultaneously Transmitting and Reflecting Reconfigurable Intelligent Surface (STAR RIS) assisted THz multiuser ultra-massive MIMO systems with information of environmental obstacles such as geometry and spatial layout taken into account for the first time. We resort to Integrated Sensing and Communication (ISAC) technique and propose a novel Microwave Coincidence Imaging (MCI) based beamforming and optimization scheme, which can adapt to environmental and user position variation. The cloud radio access networks (CRAN) structured 5G downlink signal is proposed for MCI with much enhanced resolution. We first conduct MCI to obtain 3D images of the environment including users and obstacles, based on which we calculate the optimal position of the STAR by the proposed algorithm. We further combine with channel estimation and propose a semi-passive structure of the STAR and ambiguity elimination scheme for separated channel estimation. Simulation results show that the proposed scheme increases the system performance dramatically compared to the benchmark and the case when the STAR operates with random phases. Furthermore, the proposed scheme improves the system performance on average by 14.26%, 18.35% and 60.60% for three different configurations compared to that at the 10% deviation from the optimal position of the STAR even with perfect channel state information (CSI).