經由通道正交化應用強化稀疏錯誤模型於多使用者多傳多收系統

Abstract

在第五代移動通訊系統中，使用前置編碼器（precoder）之細胞內（intercell）干擾管理技術之效能可能因為缺乏精確之通道資訊而受到影響。許多先前的文獻已經提出，在設計前置編碼器考量通道資訊不匹配可以帶來效能的增長。稀疏性增強不匹模型（Sparsity enhanced mismatch model, SEMM）之前置編碼器已經被提出，其利用多傳多收(MIMO)通道矩陣內含的稀疏性，增加目標使用者的接收能量，同時限制了對其他使用者所造成的干擾，證明能比傳統範數球不匹配模型（Norm ball mismatch model, NBMM）之前置編碼器擁有更好的效能。但利用此模型所設計的前置編碼器是一個單一使用者傳輸方案，只針對單一使用者傳輸訊號。然而同時傳輸訊號給多個使用者更符合實際通訊系統場景，為了能夠同時傳輸訊號給每個使用者，且控制使用者之間的干擾，本文將提出一個具有兩階段設計之前置編碼器，結合SEMM前置編碼器，且利用Gram-Schmidt 正交化設計第二階前置編碼器，使個別使用者通訊通道矩陣的主要係數 (Principal components, PC) 分佈不同位置，讓第一階前置編碼器在增加各個使用者接收能量的同時也限制了使用者彼此之間的干擾，稱之為透過正交化之多使用者稀疏性增強不匹配模型前置編碼器 (Multiuser “orthogonalizing” SEMM precoder, MUOSEMM precoder)。在後文會加以分析解釋 MUOSEMM 前置編碼器為何可以同時傳送訊號給多個使用者，且比傳統範數球不匹配模型之前置編碼器擁有更好的效能。雖然提出的多使用者傳輸方案可以應用在多使用者系統，但當使用者通道之間存在高度相關，亦即使用者分佈較為密集，前置編碼器的效能將被大大的削減，為了克服這個問題，文末提出了使用者排程作為解決方案: 與其他使用者通道有較高相關的使用者通道將先被排除在系統之外，使剩餘使用者能夠得到較好的效能。

The performance of precoder-based spatial intercell interference management in com-munication networks can be hindered by a lack of accurate channel state information. Pre-vious works have shown that designing precoders with channel-state information (CSI) mismatch models can bring performance improvement when CSI error is present. It has been proposed to design precoders by exploiting the opportunities provided by the dual parame-ters in the Sparsity enhanced mismatch model (SEMM), which in turn exploits the inherent sparse characteristics of multiple-input-multiple-output (MIMO) interference channels, and it was shown that the result outperforms the conventional norm ball mismatch model (NBMM) precoder. In this paper, the more advanced problem of robust interference man-agement in multiuser MIMO networks is studied. The proposed two-stage MU-MIMO pre-coding scheme aims to maximize the minimum received power of each user, subject to the constraints of the leakage interference between each other. The proposed 2nd-stage precoder aims to “orthogonalize” the channels, making the principal components (PCs) locate at dif-ferent locations of the effective channel matrices. As a result, the 1st-stage precoder allows transmitters allocate higher transmit power to targeted users without causing noticeable in-terference to others. The proposed MU-MIMO precoding scheme will be referred as the multiuser “orthogonalized” SEMM precoding, or MUOSEMM precoding. We proposed three versions of the 2nd-stage precoder; the design method and their performance analysis will be given in the text. Although the proposed MU-MIMO precoding scheme can be ap-plied on multiuser system, when channels are high correlated, the performance will degrades seriously. To cope with this issue, user scheduling was proposed: the user which mostly cor-related to other users will be removed first to ensure the MUOSEMM precoders have im-proved performance for remaining users.