雙選擇性通道中基於期望值最大化之正交分頻多工系統接收機的設計與模擬

Abstract

正交分頻多工系統(Orthogonal Frequency Division Multiplexing, OFDM)之載波間干擾(Inter-carrier Interference, ICI)效應會隨著車速提高而越益嚴重。本論文針對此問題，提出兩種遞迴式接收機來抑制載波間干擾，第一種為基於期望值最大化(Expectation-Maximization, EM)之最大相似度判定(Maximum Likelihood, ML)接收機，另一種則為加入迴旋編碼之基於期望值最大化渦輪式(Turbo-EM)接收機。首先，分析訊號載波間干擾效應的頻域模型，並結合期望值最大化演算法及載波間干擾消除來進行遞迴式的通道統計特性估測與資料偵測，以達到最大相似度判定準則，此為ML-EM接收機。然而，ML-EM接收機雖可改善使用一階等化器(One-tap Equalizer)的錯誤率，在高訊雜比(SNR)時還是會有明顯的錯誤基底(error floor)，為了解決此問題，我們提出一加入編碼的Turbo-EM接收機，此接收機使用最大後置機率判定之期望值最大化(MAP-EM)演算法，並利用雙向軟式輸出維特比演算法(Bi-directional Soft Output Viterbi Algorithm, Bi-SOVA)來提供MAP-EM所需的傳送訊號之事前機率，在通道平均值已知的情況下，Turbo-EM接收機可明顯縮減ML-EM接收機在高正規化都卜勒頻率(>0.1)時與理想ML-EM(假設傳送資料已知)間的差距，使兩者之間的效能更貼近。

In mobile environments, time-variant channels will cause inter-carrier interference (ICI) in orthogonal frequency division multiplexing (OFDM) system, thus severely degrade the system performance. This thesis proposes two expectation-maximization (EM)-based iterative receivers to deal with the ICI problem. First, an ML-EM receiver is developed to reduce the computational complexity, the main idea of which is to combine the EM algorithm with a groupwise ICI cancellation method for iterative data detection and channel estimation. The receiver effectively improves the bit error rate (BER) performance of the conventional one-tap equalizer, but the performance degradation compared with the perfect case (assuming both average channel gain and transmitted data symbol we known) is still considerable. We then investigate an EM-based turbo receiver for OFDM systems with outer channel coding, called a Turbo-EM receiver. Based on the MAP-EM receiver structure, this receiver employs a bi-directional soft-output viterbi algorithm (Bi-SOVA) to exchange soft information with a maximum a posteriori (MAP)-EM algorithm by using the turbo principle. Compared with the ML-EM receiver, our simulation results show that the Turbo-EM receiver significantly improves the BER performance, and its performance is more close to the perfect case, under relatively high normalized Doppler frequency(>0.1).