考慮LTE-A上鏈傳輸於通道估計誤差下之低複雜度最佳通道預測器設計

Abstract

本論文提出一個經由通道預測機制以降低獲取通道狀態資訊時之通訊冗餘的方法。由於真實世界的無線通道環境可看作是隨著時間而改變的連續函數，LTE上鏈傳輸中，傳送端連續發送數次時槽的訓練符號後，便停止發送，而接收端在估計了數次的通道狀態資訊後，便利用通道在時間上的相關性預測最新的通道狀態資訊。在本篇論文中，我們一併考慮通道估計誤差的影響，設計出一個最佳的通道預測機制，同時可降低傳送端及接收端的訓練冗餘，並且能夠提供和通道估計機制相近的效能。在設計出最佳預測機制後，我們亦著手分析此預測器的效能，推導出均方誤差和訊雜比的數學明確表示式。在模擬結果中可看出相較於傳統預測器，本論文所設計的最佳預測器效能更好、更接近通道估計機制效能，並且計算上比通道估計機制更為簡單。此外，模擬結果證實模擬值與本論文所推導的理論值相當吻合。

Energy efficiency is a critical demand in the design of next generation wireless communication systems such as Long Term Evolution (LTE). In this thesis, we study the problem of training overhead reduction for LTE uplink transmission, in which the Single-Carrier Frequency Division Multiple Access (SC-FDMA) modulation is adopted. Motivated by the fact that samples of real-world wireless channels are typically correlated in time, we propose to exploit such temporal correlation to develop a new channel prediction scheme for training overhead reduction. More specifically, assuming that the receiver has acquired a set of channel estimates, based on the linear minimum mean square error (LMMSE) rule, during a few training phases, we develop a LMMSE based channel prediction scheme which explicitly takes account of the effect of channel estimation errors. A closed-form formula for the optimal channel predictor is derived, and the achievable MSE performance is the analytically characterized. The achievable post-detection SNR performance when the predicted channels are employed in the LMMSE equalizer design is also studied. Computer simulations are used to illustrate the performances of the proposed scheme.