在此篇論文中，吾人考慮了以非正交多重接取技術結合感知無線電，讓沒有頻譜認 證的次要使用者可以利用主要使用者上行的時間進行傳輸，並且使用非正交接取技術有 效提升頻譜效益。 為了最佳化次要使用者下行鏈路的發射功率分配與波束成形，吾人考慮兩種設計： 極小化極大化決策找尋用戶傳輸率以及和均方誤差決策得到最大化總傳輸率。在找尋用 戶傳輸率吾人提出迫零的預編碼設計，次要使用者在下行時透過迫零波束成形解決用戶 間干擾問題並求得所能獲取的最小能量來觀測用戶的傳輸率，由於此最佳化問題並非凸 函數，所以利用半正定放寬技術(SDR)來解決問題，但此問題為複雜的共同最佳問題 (Jointly optimal)因此將問題分為兩步驟迭代法降低複雜度；第二種方法則是和均方誤差 準則在滿足對主要使用者不造成干擾且滿足次要使用者所要求最小服務品質(QoS)來觀 測系統總傳輸率，因為此最佳問題也非凸函數，因此也引入半正定放寬技術來解決問題。 同時也進行反證法驗證此和均方誤差決策與初始設計總傳輸率結果相同。 索;In this thesis, we investigate a hierarchical Non-orthogonal multiple access cellular system simultaneously shares the macrocell uplink spectrum with the microcell secondary user who is unlicensed. We also attempt the NOMA which is expected to improve the system capacity of wireless communication systems. We jointly design the allocated power and the beamforming weights for each downlink microcell user to maximize the capacity of the hierarchical NOMA cellular system. We consider two methods: (1) Using minimax decision theory to find the user rate. (2) Using the sum-MSE decision theory to find the sum rate. When we seek out user rate through Zero-forcing precoding to eliminate the interference from the other beam users and find the optimal power allocation. This problem is non-convex so we convert the aforementioned ZF user rate maximization problem into a convex semi-definite relaxation (SDR) techniques. Because this problem is a joint design non convex problem, it is difficult to jointly find the optimal solutions of power allocation and beamforming design. We divide the problem into two steps to alternatively find the optimal solutions. Second method is using sum-MSE to seek the maximization sum rate of the system, while satisfying the constraints of the interference power to the primary user, the required SINR for secondary users, and the SIC requirement for the user in the same beam. This problem is also the non-convex problem so we solved it by SDR approach. And then we proof that sum-MSE method sum rate is equal to the initial designing sum rate problem.
