多輸入多輸出單載波分頻多重存取傳輸系統中有效率之資料偵測演算法

Abstract

現代的通訊系統為了提供高品質的服務，對於高資料傳輸率與高系統效能的需求增加。在3GPP定義的長期演進(LTE)之上行鏈路，單載波分頻多重存取搭配多輸入多輸出空間多工技術(MIMO Spatial Multiplexing)可提供所需高資料傳輸率。在MIMO SC-FDMA空間多工系統中，最大概似檢測可提供最佳解，但資料維度太大導致高運算複雜度而難以實現。所以最小均方誤差與迫零常被用來當作次佳解，雖然複雜度極低，但效能卻不好。在本篇論文中，將介紹其他次佳資料偵測演算法，並提出後偵測處理輔助隨機擾動與混和型兩種較有效率之資料偵測演算法。最後，藉由電腦模擬驗證此兩種演算法在資料維度增加與高訊雜比區域，可降低運算複雜度且獲得較好的效能。

In order to provide high quality services, current communication systems require high data rates and good performance. 3GPP prescribes SC-FDMA for uplink transmission in the LTE. SC-FDMA with MIMO spatial multiplexing technology can provide high data transmission rate. In MIMO SC-FDMA spatial multiplexing systems, maximum likelihood detection can provide optimal solution, but the data dimension is too large, thereby leading to prohibitive computational complexity. So MMSE and zero forcing are often used as the sub-optimal solution. Although the computational complexity is very low, the performance is not good. In the thesis, we introduce some sub-optimal data detection algorithms. Next, we propose two efficient data detection algorithms named PDP-assisted Random Perturbation algorithm and Hybrid algorithm. Finally, we verify with computer simulation that these two algorithms can reduce the computational complexity and obtain better performance, particularly as the data dimension increases or in the high SNR region.