高維度多輸入多輸出系統下實現最佳接收機的物理層抽象技術之研究

Abstract

在第五代行動通訊系統要實現1,000倍的通道容量提升，多輸入多輸出將扮演非常重要的角色。整體網絡效能需要從鏈路級和系統級做評估。然而在多輸入多輸出系統下用戶設備的鏈路級性能難以轉換到系統級仿真（特別是對於最佳接收器）。對於移動站和基地台之間的無線電鏈路，物理層模擬器被整合在系統級仿真是不切實際，因為它需要大量的計算功率。 在本文中，對於最佳接收器我們著重於物理層性能抽象技術和開發系統級仿真鏈接到系統接口映射。在高維多輸入多輸出系統下，我們提出了以下幾種方法樹形數據結構相、改進的快速高斯變換和球形解碼來計算互信息。這些方法的結果顯示都大幅地降低計算複雜度。尤其在高階調製和高信噪比下使用球形解碼方法效果更顯著。

Multiple-input multiple output (MIMO) will play a key role to achieve the goal of 1000 times capacity enhancement for 5G systems. The network performance needs to be evaluated from both the link level and the system level as well. However, it is difficult to translate the link level performance of a user equipment (UE) receiver (especially for the optimal receiver) to system level simulations with MIMO systems. Physical layer (PHY) simulators for radio links between the mobile stations and base station is impractical to be integrated in system level simulations because it requires large computational power.

In this thesis, we focus on PHY performance abstraction techniques for the optimal receiver and develop a link-to-system interface mapping for system-level simulations. We present a tree data structure combining with methodology; improved fast Gauss transform (IFGT), and sphere decoding (SD) in computing the mutual information of the optimal receiver in high-dimension MIMO system. The results of these methods display significantly reduce the computational complexity. In particular, using the sphere decoding method works in the high-order modulation and a high signal to noise ratio is more significant.