| 摘要: | 隨者科技的進步，無人機(UAV)與智能反射板(IRS)的研究迅速發展，其相關應用將可以改善無線通訊系統訊號品質；藉由無人機高操控性、部署簡單、成本低廉等特性，對於改善過往蜂巢式網路中，基地台於城市場域中和使用者間之通道因遮蔽效應造成通訊品質低下有很大的幫助，另外智能反射板也是一新興技術，藉由布置於高樓大廈外牆，將可控制反射信號角度進而增加傳輸信號強度，進而大幅提升通道品質。蜂巢式網路中位於細胞邊緣之使用者因與基地台位置較遠，容易因為通道路徑損失影響而造成通訊品質不佳，此外其他基地台對於使用者之干擾又近一步影響其效能，為此3GPP提出一協調式多點技術(Coordinated Multi-Point, CoMP)，藉由基地台間共享通道狀態資訊，將可以藉由排程減少系統間各基地台之干擾亦或是多基地台共同傳輸相同信號以提高使用者吞吐量等，上述技術皆可對通訊系統有所幫助，因此本研究藉由將無人機搭載多天線基地台形成一多無人機協作式通訊系統，在都市條件下佈置多塊智能反射板，設計無人機軌跡、波束成形並計算反射板之反射單元角度，以達到整體系統吞吐量最大化。然而此問題為一複雜且非凸優化問題，吾人以交替優化形式拆解成三個子問題: (1) 固定無人機飛行軌跡與智能反射板反射單元角度前提下設計基地台波束成形，在有限天線數量條件下，設計波束成形使干擾傳輸信號通道處於主要傳輸信號通道之零空間以消除無人機群間之干擾；在巨量天線條件下，考慮各無人機群因天線間干擾互為獨立同分布而趨近於零，分析出波束成形之漸近解為最大比值合併(Maximum Ratio Transmission, MRT)方法。(2) 固定無人機飛行軌跡與基地台波束成形前提下設計智能反射板每一反射單元之角度，在有限天線數量條件下，提出一半正定鬆弛(Semi-Definite Relaxation, SDR)最佳化方法計算反射角度，並與現有之暴力破解法及最小平方法進行比較；進一步在巨量天線條件下，推導出反射單元之角度最佳漸近解。(3) 固定智能反射板之反射單元角度與基地台波束成形下設計無人機飛行軌跡，在有限天線數量條件下，將非凸優化問題經由鬆綁與變數變換轉換成一個凸優化問題並藉由連續凸逼近方法設計無人機飛行軌跡；在巨量天線條件下，藉由獲得的波束成形與反射角度漸近解簡化條件限制式，獲得無人機飛行軌跡漸近解。透過電腦模擬驗證效能，比較不同無人機天線數與有無智能反射板輔助對於系統吞吐量之影響，數值結果顯示智能反射板之輔助可有效改善系統吞吐量表現，此外本研究提出之基於半正定鬆弛之反射角度演算法優於最小平方演算法之系統吞吐量，亦可在略遜於暴力破解法效能下大幅降低系統複雜度。;With the advancement of technology, the research of unmanned aerial vehicles (UAV) and intelligent reflective surface (IRS) is rapidly developing, the related applications will improve the signal quality of wireless communication systems. In addition, intelligent reflective surface is also an innovative technology, which can be placed on the external wall of high buildings to control the angle of the reflected signal and increase the strength of the transmitted signal, thus significantly improving the channel quality. Users at the edge of cells in cellular networks are located far away from the base station and are suffered from poor communication quality due to channel path loss. In addition, the interference of other base stations to users further affects its performance, therefore, 3GPP proposes a Coordinated Multi-Point (CoMP) technology. By sharing the channel state information among the base stations, it is possible to reduce the interference among the base stations by scheduling or to increase the user throughput by transmitting the same signal from multiple base stations. The UAV trajectory, beamforming, and the angle of the reflecting unit of the IRS are calculated to maximize the system throughput. However, this problem is a complex and non-convex optimization problem, which we break down into three sub-problems in the form of alternating optimization: (1) Under the condition of fixed UAV flight trajectory and the angle of the reflecting unit of the intelligent reflection surface, the beamforming of the base station is designed so that the interference transmission signal channel is located in the zero space of the main transmission signal channel to eliminate the interference among the UAV groups; under the condition of massive MIMO, the interference among the UAV groups is considered to be independently distributed and tends to zero, and the asymptotic solution of beamforming is analyzed as the Maximum Ratio Combining (MRC) method. (2) The angle of each reflector of the intelligent reflection surface is designed under the fixed UAV trajectory and beamforming of the base station, and under the condition of the limited number of antennas, the Semi-Definite Relaxation (SDR) optimization method is proposed to calculate the reflection angle, also, the optimal asymptotic solution for the angle of the reflecting unit is further derived under the condition of massive MIMO. (3) The UAV trajectory is designed by fixing the reflecting unit angle of the intelligent reflection surface and the beamforming of the base station. Under the condition of a limited number of antennas, the non-convex optimization problem is transformed into a convex optimization problem by relaxation and variable transformation, and the UAV trajectory is designed by successive convex approximation methods. The asymptotic solution of the UAV trajectory is obtained by simplifying the conditional constraint with the obtained beamforming and reflection angle asymptotic solutions under massive MIMO.
The numerical results show that the intelligent reflection surface can effectively improve the system throughput performance. In addition, the reflection angle algorithm based on semidefinite relaxation and the trajectory design approach proposed in this study can effectively improve the system throughput. |
