本文提出適用於第五代行動通訊系統的使用者選擇和功率配置的演算法，應用在多輸入多輸出之正交分頻多工下鏈的非正交多重存取系統，目標為最大化使用者公平性和使用者吞吐量總和。在第五代行動通訊系統中，使用者的需求量大幅增加，如何服務更多的使用者並且更有效率的運用頻譜資源是第五代行動通訊系統中主要的議題之一，而非正交多重存取系統即是為此設計。與以往的正交存取系統不同，同一子載波上使用同一個與編碼的使用者不再只有一個，因此如何在干擾存在下的情況下配置功率更顯重要。 在被干擾影響的環境下，功率配置是一個非線性非凸函數最佳化問題。以往根據二分逼近法所提出的功率配置演算法應用在干擾存在的環境下只能找到局部最佳解，而本文將非線性非凸函數拆解成兩種不同的凸函數，改寫成全域最佳化問題，並計算在具有限制條件下的非線性非凸函數之最佳解。 模擬結果顯示本論文所提出的演算法具有較低的複雜度，能在最大化使用者公平性的前提下，提高使用者吞吐量總和。 ;In this thesis, we develop user selection and power allocation algorithm for the 5th generation mobile networks in downlink MIMO-NOMA system. We have set our sights on finding out the maximization of users’ fairness and their sum of throughput. One of the popular technologies in the 5th generation wireless systems is how to allocate spectrum resources in an efficient way. Therefore, Non-Orthogonal Multiple Access has been proposed. Because the user in a subcarrier with the same precoding vector is no longer single, assigning power with interference channel is more and more important. The power allocation is in interference environment with crosstalk. In this case, the problem is a nonlinear non-convex optimization problem not a convex problem, which has higher complexity and could not be solved by conventional water filling method directly. We design a bisection-based power allocation algorithm by rewriting a nonlinear non-convex optimization problem into two different convex functions, which enable us to find global optimal solutions under several limit factors. Based on the simulation results, the method we proposed enable us to ensure both system throughput and fairness among users with low-complexity.
