Terahertz (THz) communications have been widely considered as one of the promising technologies for future 6G wireless systems. In order to cope with the high path loss in THz systems, reconfigurable intelligent surface (RIS) composed of low-complexity reflecting elements can be deployed to generate directional beams. In order to acquire the direction of user equipment (UE) to send directional beams, the acquisition of accurate channel state information (CSI) is very important. Beam training is widely utilized to acquire the CSI. However, existing beam training schemes have not taken the wideband beam split effect into consideration, so the beam training accuracy decreases a lot in wideband scenarios. To solve this problem, in this paper, we propose an analytical beam training framework in RIS-assisted wideband THz communication systems. Specifically, we first propose a power distribution pattern (PDP) based direction estimation scheme, where the exact value of the received power is utilized to analytically calculate the UE direction. Then, we design the analytical codebook for the proposed framework based on the inherent parameters of the wideband THz system. Simulation results show that the proposed framework can achieve the near-optimal achievable rate performance with a lower beam training overhead than existing schemes.

Reconfigurable intelligent surface (RIS) is an emerging technology to improve the spectral efficiency of wireless communication systems. However, the increase of RIS elements results in the non-negligible overhead of channel estimation and channel feedback, as well as the high complexity of beam design. Therefore, how to improve the system capacity with a limited number of RIS elements becomes a challenge. In this paper, we propose a brand new irregular RIS structure to enhance the capacity of RIS-aided wireless communications. The key idea is to irregularly configure a given number of RIS elements on an enlarged surface, which provides extra degrees of freedom and spatial diversity compared with the classical regular RIS. For the proposed irregular RIS-aided communication, we then formulate the joint topology and beamforming design problem to maximize the system capacity. Accordingly, we propose a joint optimization framework with low complexity to alternately optimize the RIS topology and the corresponding beamforming design. Finally, simulation results demonstrate that the proposed irregular RIS with a limited number of RIS elements can significantly enhance the system capacity compared with the traditional regular RIS.

Hybrid precoding is an effective method for millimeter-wave (mmWave) massive multiple input multiple output (MIMO) system to reduce the number of radio frequency (RF) chains without an obvious sum-rate loss. However, the current literature on hybrid precoding considers either the high resolution (HR) phase shifters (PSs) with high power consumption or the impractical narrowband mmWave channel model. To this end, in this paper, we investigate an energy-efficient hybrid precoding scheme using one-bit PSs for the practical frequency-selective wideband mmWave massive MIMO systems. Specifically, we provide energy consumption analysis to reveal the fact that the energy consumed by the one-bit PSs is much less as compared to the HR-PSs and the array gain loss incurred by using one-bit PSs is also very small. In addition, motivated by the cross- entropy optimization (CEO) algorithm evolved for machine learning, we propose the CEO-based hybrid precoding scheme to maximize the achievable sum- rate of the considered system. The performance evaluation demonstrates that our proposed scheme can obtain the near-optimal sum-rate and considerably higher energy efficiency than some existing works.