可重構智能表面輔助通訊系統的離散相位量化位置

Abstract

在通訊環境中，如何降低傳輸功率的損耗是一個重要的議題。其中，可重構智能表面 (RIS) 是一種可行的方法。該方法主要利用其表面元件來改變入射波的相位，使得多條路徑能夠於接收端產生建設性干涉。 為了更貼近實際情形，本文主要討論離散相位系統。於此假設，對於RIS元件，我們只能選擇一些有限的相位，我們這些相位稱為離散相位量化位階。當所選擇的RIS元件相位跟最佳RIS元件相位兩者分布越接近，RIS輔助通訊系統的表現就會越優秀。前人的研究主要採用均勻離散相位量化位階來實作離散系統。因為沒有考慮到傳輸路徑跟實際硬體的限制，選出的RIS元件相位跟最佳的相位會有一定的落差，而這個部份是能夠被改善的。 參考S. Abeywickrama、R. Zhang和C. Yuen在2020年提出的AO-based 演算法，能在連續相位的RIS輔助系統中找到最佳的相位分布。這個分布可以用來判斷離散相位量化位階的優劣程度，我們稱其為理想的RIS元件相位分布。 為了降低離散RIS相位跟理想相位的落差，文章中提出一種新的方式來決定離散相位量化位階。透過將離散RIS輔助通訊系統的最佳化問題進行適當的數學運算，可以獲得理想相位的算式。再經過一些近似，找到理想相位分布的近似式。得知通道模型、路徑損耗常數、天線個數以及實用相移模型的算式後，藉由這個近似式，算出最合適的離散相位量化位階。 實驗結果顯示，以相同的系統效能來看，比起均勻的離散相位量化位階，文章中提出的離散相位量化位階更節省RIS的硬體資源。此外，系統使用基地台來運算離散相位量化位階，這些額外的運算成本不會算進RIS的成本內。上述特點有利於RIS的大量部署，能有效的降低傳輸功率成本。

Reducing transmission power loss needs to be considered in the communication system. Reconfiguring intelligent surface (RIS) is one of the possible approaches to achieve this. By adjusting the phase of incoming waves via the elements of RIS, constructive interference can occur at the receiver's location in multiple paths during transmission. To be more realistic, we mainly study the case where the adjustable phase of the elements on the RIS is discrete. In this work, discrete phases can be selected from a set of phases called discrete phase quantization levels. The more the distribution of selected phases fits the optimal distribution, the better the performance of the communication system is. Previous works mostly use uniform discrete phase quantization levels which did not consider the transmission path and hardware limitations of the actual environment. Therefore, the determined discrete phase will have a gap with the optimal RIS element phase, and we believed it can still be improved. Referring to the AO-based algorithm proposed by S. Abeywickrama, R. Zhang, and C. Yuen in 2020, the optimal RIS element phase distribution in a continuous phase system can be found. We use this distribution as a baseline to judge the quality of different discrete phase quantization levels. Here, we call it the ideal phase distribution of the RIS element. To reduce the gap between the discrete phase and the ideal phase, we propose a new method to determine discrete phase quantization levels. In the discrete RIS-aided communication system, the formula for determining the ideal phase can be obtained through the mathematical calculation of the optimization problem. By approximating this formula, we can find an approximation of the ideal phase distribution. From the channel model, path loss exponents, the number of antennas, and the formula of practical phase shift model, we can acquire a suitable non-uniform discrete phase quantization level via this approximation formula. Under the same system performance, the simulation results show that our proposed discrete phase quantization levels require fewer hardware resources on RIS than uniform discrete phase quantization levels do. Moreover, since we use base stations (BS) to compute the proposed discrete phase quantization levels, the additional computational cost is not counted in the cost of RIS. These advantages allow multiple RISs to be deployed in the transmission environment, which prominently reduces transmission power loss.