在通訊鏈結上，由於都卜勒效應和傳送端與接收端震盪器頻率的不一致，頻率偏移與相位雜訊是不可避免的。通常相位雜訊常伴隨著時序誤差同時發生，所以這些誤差在接收端時應被補償。 為了解決這些問題，我們提出自我建置模糊類神經網路決策回授等化器(SCFNN DFE)一個低複雜度的調適性非線性等化器。它包含架構和參數學習階段，以訓練SCFNN DFE。而前饋輸入向量集合的分類，與梯度坡降法皆被用在此線上學習演算法中。 模擬顯示我們提出的設計能夠改善傳統決策回授等化器在頻率偏移、相位雜訊和時序誤差所造成的估測錯誤。In communication links, a frequency offset due to Doppler effect, and a phase noise due to distorted transmission environment and imperfect oscillators exist. Phase noises usually accompanie the problem of timing error. These errors need to be compensated at the receiver to avoid a serious degradation. To solve three difficulties, we propose a self-constructing fuzzy neural network-based decision feedback equalizer (SCFNN DFE) with a online learning algorithm containing the structure and parameter learning phases. Both the feedforward input vector classification and a gradient-descent method are for the learning algorithm. Simulations show that the proposed SCFNN DFE improves the traditional DFE in the presence of estimation errors caused by frequency offset, phase noise and timing error.
