| 摘要: | 毫米波通訊在第五代（5G）行動通訊中是一項新興的候選技術，但是毫米波通訊仍然有挑戰需要面臨，例如覆蓋範圍有限，需要建置許多基地台來傳送訊號，因此會造成成本高昂、僅支援視距傳輸等。近年，學者基於上述問題，研發一種新技術名為智能反射面板 (IRS) 技術，通過大量低成本的無源反射元件，智能地重新建構無線傳輸環境，從而提高無限通信的性能。此外，由於毫米波通道造成高路徑損耗，因此需要大規模 MIMO 天線技術通過波束成形增益進行補償。然而，想要實際模擬毫米波通道大規模 MIMO 天線技術並不是容易的，其困難在於 MIMO 系統中的每個天線都需要對應到特定的射頻 (RF) 鏈。且射頻 (RF) 鏈造成高成本問題需要考量，因此，學者提出混合波束成形結構並應用在降低能量損耗和實作的成本的解決方案。假設我們給定了完美通道狀態信息 (CSI)，我們可以設計混合預編碼器和結合器，並且在毫米波通道中傳輸資料。然而，若是無法得知通道狀態信息 (CSI)，我們就需要事先對通道進行估測，因此，有許多通道估側的方法已被提出。在本文中，我們為混合架構的毫米波信道估計問題，既應用了毫米波通道的稀疏性質，並且結合壓縮感知技術，對稀疏通道進行估測，且利用投影算法將數字基帶和模擬射頻預編碼器的設計問題簡化為可以找到最優解的子優化問題。總結，本文所提出的演算法應用在毫米波大規模 MIMO 系統上，其考慮了通道估計、波束成形和IRS相移優化的解決方案，且頻譜效率接近了在完美通道狀態信息可實現的效率。;Millimeter-wave (mmWave) communication is an emerging candidate technology in the fifth generation (5G) mobile communication, but mmWave communication still has challenges to face, such as limited coverage, the need to build many base stations to transmit signals, thus incurring costs expensive, only supports line-of-sight transmission, etc. In recent years, based on the above problems, scholars have developed a new technology called intelligent reflecting surface (IRS) technology, which intelligently reconstructs the wireless transmission environment through a large number of low-cost passive reflective elements, thereby improving the performance of wireless communication. In addition, due to the high path loss caused by mmWave channels, massive MIMO antenna technology is required to compensate by beamforming gain. However, it is not easy to actually simulate massive MIMO antenna technology for mmWave channels. And the high cost problem caused by the radio frequency (RF) chain needs to be considered. Therefore, scholars have proposed a hybrid beamforming structure and applied a solution to reduce energy loss and implementation cost. Therefore, scholars have proposed a hybrid beamforming structure as a solution to reduce the cost and energy loss of implementation. Assuming we are given perfect channel state information (CSI), we can design hybrid precoders and combiners and transmit data in mmWave channels. However, if the channel state information (CSI) cannot be known, we need to estimate the channel in advance. Therefore, many channel estimation methods have been proposed. In this paper, for the mmWave channel estimation problem of hybrid architecture, we not only apply the sparse nature of mmWave channels, but also combine the compressed sensing technology to estimate the sparse channels, and use the projection algorithm to predict the digital baseband and analog radio frequency. The encoder design problem reduces to a sub-optimization problem where the optimal solution can be found. In conclusion, the algorithm proposed in this paper is applied to mmWave massive MIMO system, which considers the solutions of channel estimation, beamforming and IRS phase shift optimization, and the spectral efficiency is close to that achievable with perfect channel state information. |
