在本篇論文中,探討正交分頻多工系統於快速時變通道環境下之影響。在此環境下,吾人提出一種新的二維通道估測方法,以及低運算複雜度的載波間干擾 (intercarrier interference, ICI) 消除方法。在此兩參數:都卜勒頻移及訊框長度為高乘積的情況下,探討正交分頻多工系統基於散佈導引信號之數種時頻通道估測演算法。首先,使用遞迴最小平方法(recursive least squares, RLS)及正規化最小均方差法(normalized least mean squares, NLMS),搭配使用一組組載波資訊,可同時抑制時變通道與I/Q不平衡造成的影響。接著,在高速衰減通道環境下,吾人提出一高精確性時頻通道估測法。設計此通道估測器的關鍵乃是基於散佈式導引信號之最大似然準則方法,此估測器可即時自動追蹤通道特性,以達到最佳性能界線及高精確性二維通道估測插補效果。 最後,當都卜勒頻移及訊框長度之乘積比大於10-2時,正交分頻系統載波間之正交性將會被破壞,由於高移動性造成之載波間干擾將會造成系統性能降低。在此提出一低運算複雜度載波間干擾消除方法,藉由雙向決策回授機制來消除高速衰減通道造成之影響。模擬結果顯示此載波間干擾消除法有效地減低了載波間干擾的現象。
In this thesis, we present a novel two-dimensional (2-D) channel estimation algorithm for fast time-varying scenario and a low-complexity intercarrier interference (ICI) cancellation scheme for mobile orthogonal frequency-division multiplexing (OFDM) systems. Under the case of high product of Doppler frequency and frame duration, a number of adaptive time-frequency channel estimation algorithms for OFDM systems with scattered pilot arrangement are investigated. First, recursive least-squares (RLS) and normalized least-mean-squared (NLMS) adaptive algorithms are combined with paired-tones method for directly suppressing both of the fading channel variation and I/Q imbalance effects at the receiver end. Next, we propose a high accuracy time-frequency channel estimator for fast fading channels. The key to make the 2-D estimator successful lies the pilot-based tap-selective maximum-likelihood channel estimation, for it can adapt itself to the instantaneously channel condition and achieve an optimal estimation bound. Finally, as Doppler-frame product increases up to the level of 10-2, the orthogonality between OFDM subcarriers is destroyed, and then ICI effect due to high mobility dominates the system performance. We propose an ICI cancellation scheme by using bi-directional decision-feedback (BDDF) structure for fast fading channel mitigation. Simulation results show that the proposed BDDF-ICI cancellation scheme can not only substantially remedy the ICI effect, but also inherit low computational complexity.