| 摘要: | 本研究設計並實現一套基於軟體定義無線電（Software-Defined Radio, SDR）平台之多輸入單輸出（Multiple-Input Single-Output, MISO）雙波束成形系統，結合方向到達估計（Direction of Arrival, DoA）與序列凸逼近（Sequential Convex Approximation, SCA）技術，以達成主波束增益最大化與干擾方向抑制的雙重目標。前端訊號處理採用 MUSIC（Multiple Signal Classification）演算法進行 DoA 偵測，根據天線初始相位完成空間域至角度域的轉換，以獲得入射訊號方位。針對旁瓣方向控制，依據天線陣列幾何與相位差，利用反正切（arctangent）函數推算角度，作為干擾方向設計依據。
波束權重設計方面，本研究提出一套基於 SCA 技術的雙波束最佳化方法，針對非凸問題進行一階泰勒展開（first-order Taylor expansion）以線性化近似，達成主波功率最大化與干擾抑制目標。設計中納入總功率門檻、旁瓣抑制條件及天線振幅一致性限制（constant modulus constraint），並於目標函數中引入懲罰項以控制每根天線功率偏移程度，避免振幅過度偏離理想值，進一步提升場型穩定性與實作可行性。上述非凸約束亦透過泰勒展開逼近其下界，使整體問題可轉換為凸形式進行求解。最終所得波束權重向量與初始相位整合後，轉換為相位控制表，應用於硬體發射模組實現波束輸出。
系統實作採用兩組 1×8 線性陣列天線作為發射端，結合毫米波上下變頻模組完成射頻訊號處理，接收端則整合高指向性 Horn 天線，建構完整毫米波通訊實驗平台。為補償硬體實作造成的相位與增益誤差，系統整合天線校正機制，提升波束成形之準確性與穩定性。效能驗證規劃兩種實驗場景：場景一為基本波束成形架構，用以評估主波增益提升效果；場景二加入多組干擾訊號源，模擬複雜通訊環境下之多目標情境。根據星座圖與誤差向量幅度（Error Vector Magnitude, EVM）等效能指標分析，結果顯示所提系統在兩種場景中皆具良好表現，展現優異之方向選擇性與波束控制效能，為多波束毫米波通訊系統設計提供具體可行之架構依據。;This study presents the design and implementation of a multiple-input single-output (MISO) beamforming system based on a Software-Defined Radio (SDR) platform. The system architecture integrates Direction of Arrival (DoA) estimation and Sequential Convex Approximation (SCA) techniques to achieve dual objectives: maximizing main-lobe gain and suppressing interference from undesired directions. In the signal preprocessing stage, the system employs the Multiple Signal Classification (MUSIC) algorithm for DoA estimation. Using the initial antenna phase information, spatial-domain data are converted into angle-domain representations to identify the direction of incoming signals. For sidelobe suppression, the interference angles are derived based on antenna array geometry and phase differences, with angle calculations performed via arctangent functions.
In terms of beamforming weight design, this study proposes a dual-beam optimization approach based on Sequential Convex Approximation (SCA). To address the non-convex nature of the problem, a first-order Taylor expansion is employed to linearize and approximate the objective, aiming to maximize main lobe power while suppressing interference directions. The design incorporates a total power constraint, sidelobe suppression requirements, and constant modulus constraints to ensure amplitude consistency across antenna elements. Additionally, a penalty term is introduced in the objective function to regulate the power deviation of each antenna element, thereby preventing excessive amplitude variation and enhancing pattern stability and implementation feasibility. The non-convex constraints are also approximated from below using Taylor expansion, enabling the entire problem to be reformulated into a convex form for efficient solving. The final optimized beamforming weight vectors are integrated with initial phase calibration and converted into a phase control table, which is applied to the hardware transmission module for beam output realization.
The system is implemented using two 1×8 linear antenna arrays as the transmitter, integrated with millimeter-wave up/down-conversion modules for RF signal processing. A high-gain horn antenna is used at the receiver to establish a complete millimeter-wave communication testbed. To address hardware-induced non-idealities such as phase offsets and gain inconsistencies, the system incorporates an antenna calibration mechanism to enhance beamforming accuracy and stability. Performance evaluation is conducted under two experimental scenarios: Scenario 1 validates basic beamforming under a single-target environment, while Scenario 2 introduces multiple interference sources to simulate a complex communication environment. Based on constellation diagrams and Error Vector Magnitude (EVM) metrics, the proposed system demonstrates robust beamforming performance and effective interference suppression in both scenarios, offering a practical and flexible framework for future multi-beam millimeter-wave communication systems. |
