| 摘要: | 本論文研究一個由多功能可重構智慧反射面（Multi-Functional Reconfigurable Intelligent Surface, MF-RIS）輔助之整合式感測與通訊（Integrated Sensing and Communications, ISAC）系統。系統中基地台配備多天線以同時服務多位單天線通訊使用者，並透過額外之感測訊號提升雷達感測能力；同時引入可同時反射與透射之智慧反射面，以增強空間自由度並改善系統覆蓋彈性。為貼近實際無線環境，本研究採用具視線與非視線成分之 Rician 通道模型，並以均勻線性陣列建立通道方向性結構，推導通訊接收訊號與等效通道模型。
為同時兼顧通訊可靠度與感測準確度，本文以使用者平均平方誤差（Mean Square Error, MSE）作為通訊效能指標，並以發射波束圖誤差衡量感測端之能量聚焦能力，進而建立聯合最佳化問題。在基地台發射功率限制下，本文納入感測波束圖誤差上限與反射面硬體限制，並比較三種架構：傳統 RIS、STAR-RIS 以及具振幅控制與總能量限制之 MF-RIS，以完整描述不同反射面結構之物理可行性與功耗限制。
由於聯合設計問題涉及波束成形矩陣與反射面參數之耦合，且包含單位模數相位限制與 beampattern 四次項等非凸結構，本文採用交替最佳化（Alternating Optimization, AO）策略，將原問題分解為多個子問題交替求解，並透過連續凸逼近（Successive Convex Approximation, SCA）與一階泰勒展開將非凸項轉化為可解之凸上界。同時，STAR-RIS 之能量耦合限制以二階錐形式處理，以確保能量守恆條件。最終，本文提出一套可由 CVX 求解之 AO/SCA 聯合設計演算法，使目標函數具單調遞減特性並收斂至穩定之區域最優解，在滿足功率與硬體限制的前提下，同時達成通訊效能提升與感測波束能量聚焦之設計目標。;This paper investigates an integrated sensing and communications (ISAC) system assisted by a multi-functional reconfigurable intelligent surface (MF-RIS). The base station is equipped with multiple antennas to simultaneously serve multiple single-antenna communication users, while additional probing signals are transmitted to enhance radar sensing capability. Meanwhile, a reconfigurable intelligent surface capable of simultaneous reflection and transmission is introduced to increase spatial degrees of freedom and improve system coverage flexibility. To closely model practical wireless environments, Rician channel models with both line-of-sight (LoS) and non-line-of-sight (NLoS) components are adopted, and uniform linear arrays (ULAs) are employed to characterize the directional structure of the wireless channels, based on which the received signal model and the corresponding equivalent channels are derived.
To jointly guarantee communication reliability and sensing accuracy, the user mean square error (MSE) is adopted as the communication performance metric, while the transmit beampattern error is used to quantify the energy focusing capability in sensing. Based on these metrics, a joint optimization problem is formulated under the base station transmit power constraint, incorporating the beampattern error threshold and hardware constraints imposed by different RIS architectures. Three representative architectures are considered and compared, including conventional RIS, simultaneously transmitting and reflecting RIS (STAR-RIS), and MF-RIS with amplitude control and total energy constraints, thereby providing a comprehensive characterization of the physical feasibility and power consumption of different RIS structures.
The formulated joint design problem involves coupled beamforming matrices and RIS parameters, as well as nonconvex structures arising from unit-modulus phase constraints and high-order beampattern terms, which make the problem difficult to solve directly. To address these challenges, an alternating optimization (AO) framework is adopted to decompose the original problem into multiple subproblems that are solved iteratively. Moreover, successive convex approximation (SCA) combined with first-order Taylor expansions is employed to construct tractable convex upper bounds for the nonconvex terms. In addition, the energy coupling constraints of STAR-RIS are handled in the form of second-order cone (SOC) constraints to ensure energy conservation. As a result, an AO/SCA-based joint design algorithm is developed and can be efficiently solved using CVX. The proposed algorithm guarantees a monotonically decreasing objective value and converges to a stable locally optimal solution, achieving improved communication performance and effective sensing beampattern focusing under practical power and hardware constraints. |
