- 學位論文
近乎最佳化之大規模多輸入-多輸出系統預編碼器
An Near-Optimal Precoder for Massive-MIMO Systems
摘要
在大規模的多輸入-多輸出系統上，當訊雜比(SNR)很大時，傳統的線性預編碼方法像是Zero-forcing(ZF)有著近乎最佳化的效能表現，但對於大型矩陣來說，矩陣逆運算的複雜度通常會隨矩陣大小成立方性的增加。為了解決這個問題，我們提出了一個基於Neumann級數以及Jacobi迭代的預編碼方法，來趨近矩陣的逆運算，進而達到和傳統ZF預編碼幾乎一樣的效果。最後根據模擬結果，我們引入了曲線配適(Curve Fitting)，因應各種情況下找出最佳迭代數值的封閉表達式，並且比較其求出的數值與實際模擬結果之均方根誤差。
並列摘要
Conventional linear precoding schemes in massive multiple-input-multiple-output (MIMO) systems, such as Zero-forcing (ZF) precoding, have near-optimal performance but suffer from cubically-increased computational complexity due to the direct matrix inversion of large size. To solve this problem, we propose a modifying Neumann series and Jacobi iterative based precoding scheme to approximate the matrix inversion. The proposed precoding can approach the classical ZF precoding with negligible performance loss. According to simulation results, we conducted curve fitting to find a closed form which can provide the best iteration number for each case then we calculated the minimum root-mean-square error. Finally we also proposed a closed form for the best iteration number.