應用於V-BLAST系統下之以廣度為主做搜尋的多層排序演算法

Abstract

在多根傳送天線及多根接收天線的系統下（multi-input multi-output，簡稱為MIMO），雖然這種架構可以增加傳輸容量，但對接收端來說，如何將多根天線間彼此互相干擾的傳輸資料進行估測與解碼是主要的議題，而學者們近年來已經提出了數個解碼校能接近最佳估測的演算法（maximum-likelihood detection，簡稱為ML detection）。本篇論文著重於此系統中V-BLAST架構（Vertical Bell Laboratories Layered Space-Time）之接收端的估測，在錯誤率與複雜度的平衡間，我們結合了N-QAM多層架構搜尋的降低複雜度特性、排序QR分解的特定解碼順序與適應性調整門檻來限制搜尋路徑個數的演算法（breadth-first signal decoder，簡稱為BSIDE）[12] 而衍生出一個解碼效能逼近最佳估測且複雜度低的演算法，稱為"以廣度為主做搜尋的多層排序演算法"（multi-level sorted breadth-first signal deco- der，簡稱為MSBSIDE演算法）。經過數學分析與電腦模擬，此演算法確實能以很低的複雜度來逼近ML估測的效能，且越高階的調變越能顯現出N-QAM多層架構降低複雜度的特性，所以此演算法能在複雜度與解碼效能之間的取捨取得更佳的平衡。

The use of multi-input multi-output (MIMO) architectures promises to achieve higher transmission capacity, but how to detect and decode the transmitted signals from the interference of other antennas for the receiver is the most important topic for discussion. In recent years, a number of detecting schemes have been proposed to perform maximum-likelihood (ML) detection. In this paper, we focus on the detecting algorithm design of the Vertical Bell Laboratories Layered Space-Time (V-BLAST) receiver. Consider the tradeoff between detecting error probability and detecting comp- lexity, we combine the breadth-first signal decoder (BSIDE algorithm) [12], which can limit the number of searching paths by an adaptive threshold, with multi-level search- ing structure of the N-QAM modulation constellation and specific detecting order of sorted QR decomposition to form a novel detecting algorithm, called “multi-level sorted breadth-first signal decoder (MSBSIDE algorithm) ”, and we can let the perfor- mance of this novel algorithm approach ML detector. Through analysis and computer simulation, it is shown that the MSBSIDE algorithm has the same bit-error-rate per- formance as the conventional ML detectors while allowing significantly lower compu- tational complexity. In addition, using multi-level searching structure of the N-QAM modulation constellation for higher level modulation can reduce more computational complexity, so this novel algorithm can achieve a better performance-complexity tradeoff.