本篇論文提出的方法為，將精簡型自我組織模糊類神經網(compact self-constructing fuzzy neural network, CSFNN)用於調適性非線性波束形成(beamforming)偵測器上。CSFNN中使用精簡型架構演算法(compact self-constructing learning , CSL)，使得CSFNN具有自動擴充類神經網路架構的能力。CSL採用兩項準則，用以限制結構不必要的成長，從而降低複雜度，因此，CSFNN偵測器比起傳統偵測器更能聰明地決定類神經網路架構。而在模擬結果中也顯示出，本篇論文提出的調適性波束形成器比起傳統的線性波束形成器，能在位元錯誤率部分(bit error rate, BER)部分表現更為優越。A novel adaptive nonlinear beamforming technique is proposed based on a compact self-constructing fuzzy neural network (CSFNN) detector, which is capable of increasing the size of detector structure automatically by a compact self-constructing learning (CSL) algorithm. Moreover, this CSL algorithm adopts two evaluation criteria to limit the unnecessary growth of structure complexity, so the structure of CSFNN detector can be determined more intelligently than that of classical detectors. The simulation results show that the proposed adaptive beamforming approach provides significant performance gains over classical beamforming ones in terms of bit-error rate.
