對非正交多工存取之星座點間旋轉之最小成對距離最大化

Abstract

相較於傳統的正交多工存取(Orthogonal Multiple Access, OMA),非正交多工存取(Non-Orthogonal Multiple Access, NOMA)技術已被理論證實可達到更好的系統總傳輸率。我們在兩使用者高斯多工存取通道(Two-User Gaussian Multiple Access Channel, TU-GMAC)與星座點能量分配(Constellation Power Allocation, CPA) 模式的系統假設下，推導能達到聯合星座點(Joint Constellation)的交互訊息(Mutual Information)最大化的最佳旋轉角度，進行比較。模擬系統假設傳送端結合非正交多工存取與多層編碼(Multilevel Coding)，而於接收端則比較兩種不同接收器架構：聯合最大概度接收器(Joint Maximum-Likelihood Receiver)與循序干擾消除接收器(Successive Interference Cancellation Receiver)。我們的結果說明在中低訊雜比時，採用最小距離最大化的最佳旋轉角度，相較於傳統交互訊息最大化的最佳旋轉角度，可獲得額外效能增益。更重要的是，相較於交互訊息最大化的最佳旋轉角度隨訊雜比改變數值，且只能依賴模擬數值解，採用最小距離最大化的最佳旋轉角度與訊雜比無關且具備公式解，因此更適於實際應用。

Non-orthogonal multiple access (NOMA) technique has been shown to achieve a better system throughput than the traditional orthogonal multiple access (OMA). In this thesis, under the constellation power allocation (CPA) scheme over a two-user Gaussian multiple access channel (GMAC), the closed-form expression for the best inter-constellation points are established for many different combinations of modulations. Comparisons with the inter-constellation rotation angle that maximizes the mutual information (MI) are subsequently conducted. With a transmitter that combines multilevel coding with NOMA and a receiver that performs either joint maximum-likelihood (JML) or successive interference cancellation (SIC), our simulations indicate that the minimum-pairwise-distance-maximizing inter-constellation rotation can reach a better performance than the MI-maximizing inter-constellation rotation. More importantly, unlike the latter that varies with the signal-to-noise ratio (SNR) and can only be obtained numerically, the minimum-pairwise-distance-maximizing inter-constellation rotation does not depend on SNR and has a closed-form expression, and hence is more suitable for practical applications.