# 臺灣博碩士論文加值系統

(35.170.82.159) 您好！臺灣時間：2022/05/20 13:22

:::

### 詳目顯示

:

• 被引用:0
• 點閱:127
• 評分:
• 下載:7
• 書目收藏:0
 常數模數演算法與多重模數演算法有收斂速度慢的缺點，利用牛頓演算法雖然可以加速常數模數演算法多重模數演算法在盲蔽等化的收斂速度，但是也伴隨著大量的運算量，本篇論文主要的工作就是証明正交常數模數演算法與正交多重模數演算法可以取代牛頓常數模數演算法與牛頓多重模數演算法。正交常數模數演算法與正交多重模數演算法不僅可加速收斂速度還可使用格式雙向預估演算法得到較低的運算量。
 The Newton algorithm can be applied to the well known constant modulus algorithm (CMA) and the multimodulus algorithm (MMA) to accelerate their convergence speed for blind equalization. This work demonstrates that the Newton-based CMA and the Newton-based MMA can be closely approximated by the orthogonalized CMA (O-CMA) and the orthogonalized MMA (O-MMA), respectively. The O-CMA and O-MMA are shown to exhibit better numerical stability than their Newton-based counterparts. Moreover, the O-CMA and the O-MMA, both of which were conventionally implemented by using the matrix inversion lemma, can be implemented in a more computationally efficient manner by exploiting the QR-decomposition-based least-squares lattice (QRD-LSL) interpolation algorithm.
 Abstract (in Chinese).....................................iAbstract.................................................iiAcknowledgement.........................................iiiContents.................................................ivList of Figures...........................................v1 Introduction............................................11.1. The Baseband Equivalent Model .......................41.2. Constant Modulus Algorithm (CMA).....................61.3. Multiple Modulus Algorithm (MMA).....................91.4. Multiple Modulus Algorithm (p, q)...................102 CMA and MMA Based on Pseudo-Newton Algorithms..........122.1. Pseudo-Newton-CMA and Pseudo-Newton-MMA.............132.2. O-CMA and O-MMA Based on QRD-LSL Interpolation Algorithm................................................153 Simulation Results.....................................284 Conclusion.............................................52References...............................................55List of FiguresFig. 1.1: T/2-spaced noisy channel model..................6Fig. 2.1: Divide-and-conquer scheme for the equalizer with tap length ............................................27Fig. 3.1: A comparison of residual ISI using variousalgorithms with 4-QAM and SNR=30dB using Channel 1 with BSE.....................................................36Fig. 3.2: A comparison of residual ISI using various algorithms with 16-QAM and SNR=30dB using Channel 1 with BSE.....................................................36Fig. 3.3: A comparison of residual ISI using various algorithms with 32-QAM and SNR=30dB using Channel 1 with BSE.....................................................37Fig. 3.4: A comparison of residual ISI using various algorithms with 16-QAM and SNR=30dB using Channel 1 with BSE.....................................................37Fig. 3.5: A comparison of SER using various algorithmswith 16-QAM and SNR=30dBusing Channel 1 with BSE........38Fig. 3.6: A comparison of residual ISI using various algorithms with 16-QAM and SNR=20dB using Channel 1 with BSE.....................................................38Fig. 3.7: A comparison of SER using various algorithmswith 16-QAM and SNR=20dB using Channel 1 with BSE.......39Fig. 3.8: A comparison of residual ISI using various algorithms with 16-QAM and SNR=30dB using Channel 2with FSE................................................39Fig. 3.9: A comparison of SER using various algorithmswith 16-QAM and SNR=30dB using Channel 2 with FSE.......40Fig. 3.10: A comparison of residual ISI using various algorithms with 16-QAM and SNR=20dB using Channel 2with FSE................................................40Fig. 3.11: A comparison of SER using various algorithms with 16-QAM and SNR=20dB using Channel 3with FSE........... ....................................41Fig. 3.12: A comparison of residual ISI using various algorithms with 32-QAM and SNR=30dB using Channel 1with BSE................................................41Fig. 3.13: A comparison of SER using various algorithms with 32-QAM and SNR=30dB using Channel 1with BSE................................................42Fig. 3.14: A comparison of residual ISI using various algorithms with 32-QAM and SNR=20dB using Channel 1 with BSE.....................................................42Fig. 3.15: A comparison of SER using various algorithms with 32-QAM and SNR=20dB using Channel 1with BSE................................................43Fig. 3.16: A comparison of residual ISI using various algorithms with 32-QAM and SNR=30dB using Channel 3with FSE................................................43Fig. 3.17: A comparison of SER using various algorithms with 32-QAM and SNR=30dB using Channel 3with FSE................................................44Fig. 3.18: A comparison of residual ISI using various algorithms with 32-QAM and SNR=20dB using Channel 3with FSE................................................44Fig. 3.19: A comparison of SER using various algorithms with 32-QAM and SNR=20dB using Channel 3with FSE................................................45Fig. 3.20: A comparison of residual ISI using various algorithms with 16-QAM and SNR=30dB using Channel 3with BSE................................................45Fig. 3.21: A comparison of SER using various algorithms with 16-QAM and SNR=30dB using Channel 3 with BSE.......46Fig. 3.22: A comparison of residual ISI using various algorithms with 16-QAM and SNR=20dB using Channel 3with BSE................................................46Fig. 3.23: A comparison of SER using various algorithms with 16-QAM and SNR=20dB using Channel 3with BSE................................................47Fig. 3.24: A comparison of residual ISI using various algorithms with 32-QAM and SNR=30dB using Channel 3with BSE................................................47Fig. 3.25: A comparison of SER using various algorithms with 32-QAM and SNR=30dB using Channel 3with BSE................................................48Fig. 3.26: A comparison of residual ISI using various algorithms with 32-QAM and SNR=20dB using Channel 3with BSE................................................48Fig. 3.27: A comparison of SER using various algorithms with 32-QAM and SNR=20dB using Channel 3 with BSE.....................................................49Fig. 3.28: A comparison of ISI using various algorithms with 16-QAM and SNR=30dB using Channel 2 with BSE.....................................................49Fig. 3.29: A comparison of SER using various algorithms with 16-QAM and SNR=30dB using Channel 2 with BSE.....................................................50Fig. 3.30: A comparison of ISI using various algorithms with 32-QAM and SNR=30dB using Channel 2 with BSE.....................................................50Fig. 3.31: A comparison of ISI using various algorithms with 32-QAM and SNR=30dB using Channel 2 with BSE.....................................................51