臺灣博碩士論文加值系統

本論文主要探討，在可加性白色高斯雜訊及衰減通道下，設計一鎖幅迴路分離系統將主要訊號和次要訊號相互分離出來，再結合自適應線性神經元網路濾波器，可以更有效率地消除可加性雜訊信號和衰減環境下的共通道信號分離的問題。
自適應線性神經元網路濾波器是將延遲線路與自適應線性神經元網路結合起來的濾波器，以最小均方演算法(或稱Widrow-Hoff的學習規則)來估測輸出與期望訊號間的誤差，過程中藉由最陡坡降演算法來調整加權值及偏重值，找出最佳權重係數與最小均方誤差值。
模擬結果顯示出鎖幅迴路分離系統在不同信號雜訊比的環境下，結合自適應線性神經元網路濾波器能比結合傳統式自適應性濾波器有更好的信號分離效果。

In this thesis, we designed an Amplitude-Locked Loop (ALL) separation system for separating the dominant and sub-dominant signals under additive white Gaussian noise (AWGN) and fading channel. Then, the ALL system combined with an Adaptive Linear Neuron (ADALINE) network filter that can eliminate the noise interference and separate the co-channel signals more efficiently.
The ADALINE network filter is that ADALINE network combined with a tapped delay line (TDL). It is adopted to estimate the error value between the filter estimation output signal and the desired signal by using the LMS algorithm (or Widrow-Hoff learning rule) based on an approximate steepest descent procedure. The steepest descent gradient algorithm is proposed for finding the optimal weight coefficient and the minimum mean square error (MMSE).
The simulation results are demonstrated that the ALL separation system combined with ADALINE network filter has the better signal separation performance than combined with conventional adaptive filter in different signal to noise ratio (SNR) environment.

Abstract in Chinese i
Abstract ii
Acknowledgements iii
Contents iv
List of Figures vi
List of Tables viii
Abbreviations ix
Chapter 1 Introduction 1
1.1 Introduction 1
1.2 Motivation 1
1.3 Organization of this thesis 2
Chapter 2 Principle of ADALINE Filter 4
2.1 Analysis of ADALINE filter 4
2.1.1 ADALINE network 4
2.1.2 ADALINE filter model 5
2.2 Widrow-Hoff learning rule 7
2.2.1 Performance surface and performance optimization 7
2.2.2 LMS (Least Mean Square) algorithm 12
2.2.3 Analysis of convergence 14
Chapter 3 Analysis of Proposed Separation System 17
3.1 Background 17
3.2 Principle of ALL system 18
3.2.1 PLL (Phase-Locked Loop) model 18
3.2.2 ALL (Amplitude-Locked Loop) model 21
3.3 Analysis of proposed model 24
3.3.1 Co-channel interference (CCI) FM signals 24
3.3.2 Proposed separation model 29
Chapter 4 Simulated Results 35
4.1 Simulated results of ALL separation system 35
4.1.1 Simulated FM signal under interference channel 35
4.1.2 CCI FM signal separated with ALL cancellation system 39
4.2. Simulated results of proposed system 43
4.2.1 CCI signal separated by ALL system with ADALINE filter43
4.2.2 Simulated comparison for different filters 44
4.3. Simulated comparison under different factors 46
4.3.1 Performance comparison with different SNR 46
4.3.2 Simulated comparison of convergence and the performance with training 51
4.4 Simulated performance for co-channel speech signals 55
Chapter 5 Conclusions 58
References 60
Appendix 63
List of Publications 68
Autobiography 69

[1]Yu. Hua. Gu and W. M. G. van Bokhoven,“ Frequency bin nonlinear LMS adaptive noise canceller and its application to co-channel speech noise reduction”, 1991, pp. 2822-2825.
[2]J. Benesty and Y. Huang, “Adaptive signal processing”, Springer, 2003, pp. 151-164.
[3]Simon Haykin, “Adaptive filter theory”, Prentice-Hall, Inc. 3th Edition, 1996, pp.41-49.
[4]G. J. Jong, T. J. Su, and T. J. Moir, “The performance of the amplitude-locked loop with co-channel interference”, electronics letters, 1998, vol. 34, no. 8, pp. 719-720.
[5]B. Widrow and M. E. Haft, “Adaptive switching circuits”, in IRE WESCON Conv. Rec. 1960, vol. 4, pp. 96-104.
[6]Martin T. Hagan, Howard B. Demuth, Mark H. Beale. “Neural Network Design”, ISBN: 0-9717321-0-8, pp.10.1- 10.13.
[7]Halil Yigit Adnan Kavak and H. Metin Ertunc “Using Autoregressive and Adaline Neural Network Modeling to Improve Downlink Performance of Smart Antennas”, IEEE International Conference on Mechatronics 3-5 June, 2004, pp.165 -170.
[8]Elshatshat, R.; Kazerani, M.; Salama, M.M.A.” ADALINE-based controller for active power-line conditioners”, IEEE Transmission and Distribution Conference, 11-16 April 1999, vol. 2, pp.566 - 571.
[9]Bernard Widrow and Max Kamenetsky “On the Statistical Efficiency of the LMS Family of Adaptive Algorithms” Neural Networks, Proceedings of the International Joint Conference, 20-24 July 2003, vol. 4, pp.2872 - 2880
[10]G. Cauwenberghs, “Adaptation, Learning and Storage in Analog VLSI,” in Proceedings of the Ninth Annual IEEE International ASIC Conference, Rochester, NY, 23-27 Sept. 1996, pp. 273-278.
[11]Martin T. Hagan, Howard B. Demuth, Mark H. Beale. “Neural Network Design” ISBN: 0-9717321-0-8, pp.8.1- 8.2
[12]B.Widrow and S. D. Steams. Adaptive Signal Processing, Prentice Hall, Upper Saddle River, NJ. 1985.
[13]Jose C. Principe, Neil R. Euliano , W. Curt Lefebvre “Neural and Adaptive Systems: Fundamentals Through Simulations”, Copyright 2000 © John Wiley & Sons, Inc, pp.17- 24
[14]W.L.Brogan, “Modern Contrall Theory” 3rd Ed.,Englewood Cliffs, NJ: Prentice-Hall. ISBN: 0135897637
[15]G. J. Jong, T. J. Moir, A. M. Pettigrew and T. J. Su, “Improvement of FM demodulator with co-channel FM interference”, electronics letters, 1999, vol. 35, no. 20, pp. 1758-1759
[16]John E. Ohlson, “Phased-locked loop operation in the presence of impulsive and Gaussian noise”, IEEE transactions on communications, Sept. 1973, vol. com-21, no.9, pp. 991-996
[17]T. J. Moir and A. M. Pettigrew, “Inband noise cancelling in FM systems: the white noise case”, electronics letters 23rd, ,April 1992,vol. 28 no.9, pp. 814-815
[18]G. J. Jong, T. J. Su, and T. J. Moir, “The simulation of the amplitude-locked loop with co-channel Interference using Matlab”, iasted international conference, applied modeling and simulation (AMS’ 99) Cairns, Australia, Sept. 1999
[19]G. J. Jong, T. J. Moir, and A. M. Pettigrew, “The high performance FM demodulator using the amplitude-locked loop with co-channel interference”, international symposium on communication (ISCOM), Kaohsiung, Taiwan, 1999, pp. 284-288
[20]Gwo-Jia Jong, Ci-Fang Syu, and Ming-Hsun Hsieh, ” The System Model For The Separation Of Multi-channel FM signals Using Adaptive FIR Filter“, IEEE International Conference on Signal Processing (ICSP’2004), E-05, Beijing, China.
[21]Simon Haykin, “communication systems”, John Wiley & Sons, Inc.4th Edition, pp.130-149
[22]T. J. Moir, “Analysis of an amplitude-locked loop,” Electronics Letters, vol. 31, Issue 9, 27 April 1995, pp. 694 - 695.