由於非線性偵測器的訊號分類能力勝過傳統的偵測器，所以很多非線性偵測器（如：輻射基底網路(radial basis function, RBF) 和模糊類神經網路(fuzzy neural network, FNN)）已被應用至通道等化與多根天線系統等。典型RBF偵測器可利用叢聚演算法（如：敵人懲罰競爭學習(rival penalized competitive learning, RPCL)）去訓練其參數。但RPCL常伴隨掉落至局部解與緩慢收斂等問題，因此本論文第二章便提出一個新穎的密度評估(density-evaluated, DE) 機制去改進RPCL的演算過程。所提出的DE機制會計算每個中心點的資料密度，並藉由資料密度來評估是否需要刪除所對應的中心點，使RPCL可達到快速收斂的結果，我們會使用一個非線性通道模型去展示所提出的DE-RPCL演算法。另外，RBF偵測器也已成功應用至決策迴授等化器(decision feedback equalizer, DFE)，但RBF的隱藏節點個數會隨著等化器階數或通道階數的增加而呈指數增加，這會增加RBF之複雜度，所以本論文第三章提出一個新穎快速自我建構模糊類神經網路(fast self-constructing FNN, FSCFNN)決策迴授等化器，其由多個FSCFNN偵測器所組成，但每次偵測僅啟動一個FSCFNN。由於FSCFNN的隱藏節點個數會藉由設定條件去限制其大小，所以FSCFNN DFE可控制在較RBF DFE還要小的複雜度。此外，第四章我們也提出將SCFNN偵測器加入至多根天線系統，並藉由陣列輸入訊號空間的對稱特性，提出所謂的簡化型對稱自我建構模糊類神經網路(reduced symmetric SCFNN, RS-SCFNN)波束形成器。模擬結果顯示RS-SCFNN能大幅勝過傳統調適性波束形成器，且RS-SCFNN能根據現在的通訊環境彈性且自動地決定隱藏節點的個數。

Due to the better classification capability than the traditional linear detectors, many nonlinear detectors, such as radial basis function (RBF) and fuzzy neural network (FNN), have been applied to channel equalization and multi-antenna systems. Classically, an RBF detector is trained with a clustering algorithm like rival penalized competitive learning (RPCL). However, RPCL and its improved versions are always accompanied by problems of falling in local optima and slow learning speed. Thus, the first part of this dissertation focuses on proposing a novel mechanism to prune the RPCL’s structure directly by evaluating the data density of each center unit. An ordinary channel model is simulated in this part to demonstrate the proposed clustering. The number of hidden nodes of the RBF decision feedback equalizer (DFE) can be obtained from one or two pre-known information, i.e., equalizer order and channel order. However, if the equalizer order or channel order increases, the number of hidden nodes in RBF DFE grows exponentially, so do the computation and hardware complexity. In the second part of this dissertation, a novel DFE is thus proposed by using a fast self-constructing FNN (FSCFNN) detector. The proposed DFE structure is composed of several FSCFNNs, each of which corresponding to one feedback input vector. Because the feedback input vector occurs independently, only one FSCFNN detector is activated to decide the estimated symbol. Specially, all of the hidden nodes in each FSCFNN detector are flexibly determined by the proposed learning algorithm, and can be restricted to a low amount due to setting conditions to constrain the size of structure. Therefore, the proposed FSCFNN DFE results in less complexity compared to RBF DFE. Moreover, in this dissertation, we also propose to incorporate an SCFNN-related detector into multi-antenna systems with the aid of a symmetric property of array input signal space. This novel adaptive beamformer is called reduced symmetric SCFNN (RS-SCFNN) beamformer. The simulation results are done in the rank-deficient multi-antenna systems and have shown that the adaptive RS-SCFNN beamformer extremely outperforms the classical adaptive beamformers. Besides, the proposed SCFNN-related adaptive beamformers can flexibly and automatically determine different numbers of hidden nodes for various signal-to-noise (SNR) environments, but the RBF-based adaptive beamformer must assign hidden node’s numbers as a fix constant for various SNR environments before learning.

[1] S. Haykin, Adaptive filter theory (4th Edition), Prentice Hall, 2002.
[2] T.S. Rappaport, Wireless communications: principles and practice (2nd Edition), Prentice Hall, 2002.
[3] J. Litva, T.K.Y. Lo, Digital beamforming in wireless communications, Artech House, 1996.
[4] Y. Gong, X. Hong, “OFDM joint data detection and phase noise cancellation based on minimum mean square prediction error,” Signal Process., vol. 89, pp. 502-509, 2009.
[5] M.Y. Alias, A.K. Samingan, S. Chen, L. Hanzo, “Multiple antenna aided OFDM employing minimum bit error rate multiuser detection,” Electron. Lett., vol. 39, no. 24, pp. 1769-1770, 2003.
[6] S. Chen, N.N. Ahmad, L. Hanzo, “Adaptive minimum bit error rate beamforming,” IEEE Trans. Wirel. Commun., vol. 4, no.2, pp. 341-348, 2005.
[7] J. Li, G. Wei, F. Chen, “On minimum-BER linear multiuser detection for DS-CDMA channels,” IEEE Trans. Signal Process., vol. 55, no.3, pp. 1093-1103, 2007.
[8] T.A. Samir, S. Elnoubi, A. Elnashar, “Block-Shannon minimum bit error rate beamforming,” IEEE Trans. Veh. Technol., vol. 57, no.5, pp. 2981-2990, 2008.
[9] W. Yao, S. Chen, S. Tan, L. Hanzo, “Minimum bit error rate multiuser Transmission designs using particle swarm optimisation,” IEEE Trans. Wirel. Commun., vol. 8, no.10, pp. 5012-5017, 2009.
[10] S. Gollamudi, S. Nagaraj, S. Kapoor, Y.F. Huang, “Set-membership filtering and a set-membership normalized LMS algorithm with an adaptive step size,” IEEE Signal Process. Lett., vol. 5, no. 5, pp. 111-114, 1998.
[11] Y.C. Liang, F.P.C. Chin, “Coherent LMS algorithms,” IEEE Commun. Lett., vol. 4, no. 3, pp. 92-94, 2000.
[12] S. Choi, T.W. Lee, D. Hong, “Adaptive error-constrained method for LMS algorithms and applications,” Signal Process., vol. 85, pp. 1875-1897, 2005.
[13] E.F. Harrington, “A BPSK decision-feedback equalization method robust to phase and timing errors,” IEEE Signal Process. Lett., vol. 12, pp. 313-316, 2005.
[14] S. Chen, B. Mulgrew, P.M. Grant, “A clustering technique for digital communications channel equalization using radial basis function networks,” IEEE Trans. Neural Netw., vol. 4, no. 4, pp. 570-579, 1993.
[15] J. Montalvão, B. Dorizzi, J. Cesar M. Mota, “Why use Bayesian equalization based on finite data blocks,” Signal Process., vol. 81, pp. 137-147, 2001.
[16] S. Chen, B. Mulgrew, S. McLaughlin, “Adaptive Bayesian equalizer with decision feedback,” IEEE Trans. Signal Process., vol. 41, no. 9, pp. 2918-2927, 1993.
[17] S. Chen, S. McLaughlin, B. Mulgrew, P.M. Grant, “Bayesian decision feedback equalizer for overcoming co-channel interference,” IEE Proc.-Commun., vol. 143, no. 4, pp. 219-225, 1996.
[18] S. Chen, L. Hanzo, A. Wolfgang, “Kernel-based nonlinear beamforming construction using orthogonal forward selection with the fisher ratio class separability measure,” IEEE Signal Process. Lett., vol. 11, no. 5, pp. 478-481, 2004.
[19] S. Chen, L. Hanzo, A. Wolfgang, “Nonlinear multiantenna detection methods,” EURASIP J Appl. Signal Process., vol. 9, pp. 1225-1237, 2004.
[20] J. Lee, C. Beach, N. Tepedelenlioglu, “A practical radial basis function equalizer,” IEEE Trans. Neural Netw., vol. 10, pp. 450-455, 1999.
[21] P.C. Kumar, P. Saratchandran, N. Sundararajan, “Minimal radial basis function neural networks for nonlinear channel equalization,” IEE Proc.-Vis. Image Signal Process., vol. 147, no. 5, pp. 428-435, 2000.
[22] M.S. Yee, B.L. Yeap, L. Hanzo, “Radial basis function-assisted turbo equalization,” IEEE Trans. Commun., vol. 51, pp. 664-675, 2003.
[23] A. Wolfgang, S. Chen, L. Hanzo, “Radial basis function network assisted space-time equalisation for dispersive fading environments,” Electron. Lett., vol. 40, no. 16, pp. 1006-1007, 2004.
[24] J.B. MacQueen, “Some methods for classification and analysis of multivariate observations,” Proceedings of the 5th Berkeley Symposium on Mathematics Statistics and Probability, Berkeley, U.S.A., pp. 281-297, 1967.
[25] R. Assaf, S.E. Assad, Y. Harkouss, “Adaptive equalization for digital channels RBF neural network,” Proceedings of the European Conference on Wireless Technology, Paris, France, pp. 347-350, 2005.
[26] R. Assaf, S.E. Assad, Y. Harkouss, “Adaptive equalization of nonlinear time varying-channel using radial basis network,” Proceedings of the 2006 International Conference on Information and Communication Technologies, Damascus, Syria, pp. 1866-1871, 2006.
[27] L. Xu, A. Krzyzak, E. Oja, “Rival penalized competitive learning for clustering analysis, RBF net, and curve detection,” IEEE Trans. Neural Netw., vol. 4, no. 4, pp. 636-649, 1993.
[28] Y.M. Cheung, “On rival penalization controlled competitive learning for clustering with automatic cluster number selection,” IEEE Trans. Knowl. Data Eng., vol. 17, no. 11, pp. 1583-1588, 2005.
[29] J. Ma, T. Wang, “A cost-function approach to rival penalized competitive learning (RPCL),” IEEE Trans. Syst. Man Cybern. Part B-Cybern., vol. 36, no. 4, pp. 722-737, 2006.
[30] S. Chen, T. Mei, M. Luo, H. Liang, “Study on a new RPCCL clustering algorithm,” Proceedings of the 2007 IEEE International Conference on Mechatronics and Automation, Harbin, China, pp. 299-303, 2007.
[31] X. Qiao, G. Ji, H. Zheng, “An improved rival penalized competitive learning algorithm based on fractal dimension of algae image,” Proceedings of the 2008 IEEE Control and Decision Conference, Yantai, China, pp. 199-202, 2008.
[32] S. Siu, G.J. Gibson, C.F.N. Cowan, “Decision feedback equalization using neural network structures and performance comparison with standard architecture,” IEE Proc.-Commun., vol. 137, pp. 221-225, 1990.
[33] J. Coloma, R.A. Carrasco, “MLP equaliser for frequency selective time-varying channels,” Electron. Lett., vol. 30, pp. 503-504, 1994.
[34] C.H. Chang, S. Siu, C.H. Wei, “Decision feedback equalization using complex backpropagation algorithm,” Proceedings of 1997 IEEE International Symposium on Circuits and Systems, Hong Kong, China, pp. 589-592, 1997.
[35] S.S. Yang, C.L. Ho, C.M. Lee, “HBP: improvement in BP algorithm for an adaptive MLP decision feedback equalizer,” IEEE Trans. Circuits Syst. II-Express Briefs, vol. 53, no. 3, pp. 240-244, 2006.
[36] S. Siu, S.S. Yang, C.M. Lee, C.L. Ho, “Improving the Back-propagation algorithm using evolutionary strategy,” IEEE Trans. Circuits Syst. II-Express Briefs, vol. 54, no. 2, pp. 171-175, 2007.
[37] K. Mahmood, A. Zidouri, A. Zerquine, “Performance analysis of a RLS-based MLP-DFE in time-invariant and time-varying channels,” Digit. Signal Prog., vol. 18, no. 3, pp. 307-320, 2008.
[38] S.S. Yang, S. Siu, C.L. Ho, “Analysis of the initial values in split-complex backpropagation algorithm,” IEEE Trans. Neural Netw., vol. 19, pp. 1564-1573, 2008.
[39] J.S.R. Jang, C.T. Sun, E. Mizutani, Neuro-fuzzy and soft computing - a computational approach to learning and machine intelligence, Prentice Hall, 1997.
[40] C.H. Lee, Y.C. Lin, “An adaptive neuro-fuzzy filter design via periodic fuzzy neural network,” Signal Process., vol. 85, pp. 401-411, 2005.
[41] S. Siu, C.L. Ho, C.M. Lee, “TSK-based decision feedback equalizer using an evolutionary algorithm applied to QAM communication systems,” IEEE Trans. Circuits Syst. II-Express Briefs, vol. 52, pp. 596-600, 2005.
[42] F.J. Lin, C.H. Lin, P.H. Shen, “Self-constructing fuzzy neural network speed controller for permanent-magnet synchronous motor drive,” IEEE Trans. Fuzzy Syst., vol. 9, no. 5, pp. 751-759, 2001.
[43] W.D. Weng, R.C. Lin, C.T. Hsueh, “The design of an SCFNN based nonlinear channel equalizer,” J. Inf. Sci. Eng., vol. 21, pp. 695-709, 2005.
[44] R.C. Lin, W.D. Weng, C.T. Hsueh, “Design of an SCRFNN-based nonlinear channel equaliser,” IEE Proc.-Commun., vol. 152, no. 6, pp. 771-779, 2005.
[45] A. Bateman, A. Bateman, Digital communications: design for the real world, Addison Wesley, 1999.
[46] M. Martínez-Ramón, J.L. Rojo-Álvarez, G. Camps-Valls, C.G. Christodoulou, “Kernel antenna array processing,” IEEE Trans. Antennas Propag., vol. 55, no. 3, pp. 642-650, 2007.
[47] S. Chen, A. Wolfgang, C.J. Harris, L. Hanzo, “Adaptive nonlinear least bit error-rate detection for symmetrical RBF beamforming,” Neural Netw., vol. 21, pp. 358-367, 2008.
[48] S. Chen, A. Wolfgang, C.J. Harris, L. Hanzo, “Symmetric complex-valued RBF receiver for multiple-antenna-aided wireless systems,” IEEE Trans. Neural Netw., vol. 19, no. 9, pp. 1657-1663, 2008.
[49] Q. Liang, J.M. Mendel, “Overcoming time-varying co-channel interference using type-2 fuzzy adaptive filter,” IEEE Trans. Circuits Syst. II-Express Briefs, vol. 47, pp. 1419-1428, 2000.
[50] Q. Liang, J.M. Mendel, “Equalization of nonlinear time-varying channels using type-2 fuzzy adaptive filters,” IEEE Trans. Fuzzy Syst., vol. 8, no. 5, pp. 551-563, 2000.
[51] Y.J. Chang, C.L. Ho, “Improving the BP algorithm using RPCL for FNN-based adaptive equalizers,” Proceedings of 2008 National Symposium on Telecommunications, Yunlin, Taiwan, pp. 1385-1388, 2008.
[52] Y.J. Chang, C.L. Ho, “SOFNN-based equalization using rival penalized controlled competitive learning for time-varying environments,” Proceedings of 2009 International Conference on Wireless Communication and Signal Processing, Nanjian, China, 2009.
[53] S. Han, I. Lee, W. Pedrycz, “Modified fuzzy c-means and Bayesian equalizer for nonlinear blind channel,” Appl. Soft. Comput., vol. 9, pp. 1090-1096, 2009.
[54] J. Choi, A.C. de C. Lima, S. Haykin, “Kalman filter-trained recurrent neural equalizers for time-varying channels,” IEEE Trans. Commun., vol. 53, no. 3, pp. 472-480, 2005.
[55] C.F.N. Cowan, S. Semnani, “Time-variant equalization using a novel nonlinear adaptive structure,” Int. J. Adapt. Control Signal Process., vol. 12, no. 2, pp. 195-206, 1998.
[56] R. Kohno, “Spatial and temporal communication theory using adaptive antenna array,” IEEE Pers. Commun., vol. 5, no. 1, pp. 28-35, 1998.
[57] O.A.D. Ju′nior, A.D.D. Neto, W. Mata, “Determination of multiple direction of arrival in antennas arrays with radial basis functions,” Neurocomputing, vol. 70, pp. 55-61, 2006.
[58] F. Ling, J.G. Proakis, “Lattice decision feedback equalizers and their application to fading dispersive channels,” Proceedings of 1983 International Conference on Communications, Boston, U.S.A., pp. C8.2.1–C8.2.5, June 1983.
[59] 羅振原, 翁萬德, “行動無線通訊衰減通道模擬器之研究,” 國立雲林科技大學電機工程系碩士論文, June 2001.