本文提出使用者選擇與功率配置演算法，目標為考慮使用者吞吐量總和最大化與公平性之間的補償，應用在多輸入多輸出正交分頻多工下鏈系統，並且使用最小均方誤差預編碼。因為找尋最佳使用者之集合計算複雜度過於龐大，所以我們使用一些數學簡化的程序來減少複雜度。然而干擾存在下之功率配置方式不再與傳統注水式功率配置演算法相同。在這個議題下，問題將不再是凸函數之問題，而是轉變成非線性非凸函數最佳化問題。我們根據二分逼近法提出功率配置演算法處理干擾存在下功率配置的問題只能找到局部最佳解。因此，我們使用全域最佳化方法計算帶有限制之下非線性非凸函數之最佳化問題。其方法將非線性非凸函數取代為兩種不同的凸函數，並計算全域最佳解。由複雜度與電腦模擬分析，我們可以知道提出次佳演算法逼近最佳解，並且有更少的複雜度，同時使用者吞吐量與公平性之間的補償也被考慮在其中。 In this thesis, the user selection and power allocation algorithm is proposed for downlink MIMO-OFDM system using MMSE precoding, where the objective considers balancing between the maximization of the sum of users’ throughput and their fairness. Since finding the optimal user subset has very high computational burden, we use some mathematical simplification processes to reduce the scheduling complexity. However, the power allocation is not in interference-free environment as in the conventional water filling method but in the presence of crosstalk. In this scenario, the problem under consideration is not a convex problem but a nonlinear non-convex optimization problem, which is difficult to solve. We propose a power allocation method based on the bisection strategy to overcome this problem with interference-aware capability of finding local optimal solutions. In addition, we use global optimization techniques to compute global optima of the constrained non-convex nonlinear optimization problems. Its objective function is replaced by a difference of two convex functions. The computer simulation results show that the proposed sub-optimal scheme is close to that of the optimal solution with a less complexity and the tradeoff between system throughput and fairness among users is considered.
