針對MIMO頻率選擇性信道提出了一種分數間隔的判決反饋均衡器，其優點是既對時間同步誤差不敏感，又能在消除碼間干擾的同時不引入噪聲增益。同時比較了分別采用最小均方算法和恒模算法的分數間隔判決反饋均衡器，并通過仿真，分析了這兩種算法的優缺點。在此基礎上提出了一種改進的最小均方算法和恒模算法相結合的算法，改進后需要的訓練序列為LMS算法的1/3，其結果是誤碼率比恒模算法有明顯的提高。
關 鍵 詞 MIMO頻率選擇性信道; 分數間隔均衡; 判決反饋均衡; 最小均方算法; 恒模算法

Abstract This article researches the combination of fractionally spaced equalizer (FSE) and decision feedback equalizer(DFE) applied in MIMO frequency selective channels. This kind of equalizer has low sensitivity to the receiver’s sampling phase, and it can eliminate the ISI without adding noise gain. This paper analyses and compares the FSE-DFE applying two adaptive algorithms: least mean squares (LMS) algorithm and constant modulus algorithm (CMA). Through the simulation results, their respective advantages and disadvantages are presented. Based on these algorithms, a new hybrid algorithm possessing the advantages of LMS and CMA is proposed. This algorithm requires only a few training sequences and performs better than CMA.
Key words MIMO frequency selective channels; fractionally spaced equalizer; decision feedback equalizer; least mean squares algorithm; constant modulus algorithm