無線中繼網路之用以減少訓練冗餘的最佳線性通道預測：演算法設計與效能分析

Abstract

在現今的無線中繼網路中，有效率的使用系統能量已經成為一個相當重要的設計考量與依據。從實體層的觀點出發，發展一套以較少的信令冗餘來獲得通道狀態資訊的機制一直受到了相當多的關注。在本篇論文中，我們考慮的環境是一個採用＂預先最大比率結合＂來增強通道鏈可靠度的無線中繼網路，並提出一個基於線性通道預測的兩階段式的通道狀態資訊獲得機制。我們假設無線通道是平坦衰弱的，且在時間上是存在相關性的。在提出的方法中，我們在訓練階段(階段I)時，採用了基於線性最小均方誤差準則的估計法，來估計通道增益。接著，在預測階段(階段II)時，我們利用在階段I所收集的通道估計結果，透過＂線性最小均方誤差通道預測器＂，來預測通道增益。其中，我們考量了通道估計誤差的影響來設計我們的通道預測器，並推導出最佳預測器完整的數學表示式。我們接著分析了這二種階段在接收端的訊雜比表現，並推導出訊雜比的數學公式，最後再透過電腦模擬來驗證我們的分析結果。

Energy-efficiency has become a critical design concern in modern wireless relay networks. From a physical-layer perspective, development of channel-state-information (CSI) acquisition protocols with low signaling overheads has received considerable attention. In this thesis, we consider a wireless relay network employing pre maximum ratio combining (pre-MRC) for link reliability enhancement, and propose a two-phase CSI acquisition scheme based on the linear channel prediction principle. We assume that the wireless channels are flat-fading and temporally correlated. During the training phase (Phase I) of the proposed method, the channel taps are estimated using the linear minimum mean square error (LMMSE) criterion. Then, in the channel prediction phase (Phase II), the channels are predicted by using the LMMSE channel predictor with the estimated channels in Phase I as the predictor inputs. The considered optimal predictors explicitly take into account the channel estimation errors, and are derived in closed-form. Exact SNR analyses are then conducted to obtain closed-form formulas for the receive SNR at the destination. Computer simulations are used to validate our analysis.