在分碼長多工存取架構下以天線陣列擷取展頻訊號

Abstract

在本論文中，吾人提出一分碼長多工存取架構之波束合成法，用來增加傳 統分碼多工存取架構之容量。此架構將不再限制所有使用者之展頻訊號碼 重復率為單一值，從而利用不同碼重復率所具有之週期穩態特性來區隔不 同的使用者。如此一來，只要利用天線陣列先在空間上區分不同群組的使 用者，就可以允許更多的使用者在相同的頻帶通訊而不會互相干擾。首先 ，吾人利用一名為自同調強度回復演算法來找出特定群組之訊號子空間， 再透過一成本函數將這些訊號特徵向量合成能同時接收特定群組內所有使 用者之單一加權向量。此成本函數乃藉由最小化波束合成器之輸出功率， 並限制合成向量之首位元素為一定值。為了避免此波束合成器在高信號雜 訊比時產生目標信號消除之效應，吾人在自相關矩陣上加入一適當大小之 白色假雜訊功率以可改善此一現象。此外，為了能更正確地鑑別目標信號 群，吾人亦提出了使用兩個不同週期特徵之兩級波束合成器架構，除可達 到增強波束合成器信號選擇能力之目的外，亦大幅降低所需之運算量。最 後，吾人以電腦模擬的結果來驗證以上所提出的各種方法。 A novel beamforming scheme for increasing the capacity of the code division (CDMA) multiple access systems is presented. The scheme is developed based on the cyclostationarity of the direct sequence binary phase shift keying (DS-BPSK) signaling, especially for chip rate and code repetition frequency (CRF). The beamformer identifies each user group by its associated CRF in a particular spectral band. An eigen-based self-coherence restoral (SCORE) algorithm is employed to construct the signal subspace corresponding to the direction vectors of multiple signals of interest (SOI's). A grouping of these signal eigenvectors is then performed which produces a single weight vector whose pattern encompasses all of the SOI's simultaneously. The grouping is done so as to minimize the output noise power of the beamformer subject to a constraint of unit leading component of the weight vector. To remedy the performance degradation due to desired signal cancellation under a high signal-to-noise ratio level, we suggest adding the regular autocorrelation matrix a white pseudo noise component. Also, a two-stage scheme is proposed to decompose the beamforming process into two individual stages which enable one to select SOI's with two specified cyclic features. This helps to enhance the selectivity of the beamformer. Finally, a complete set of numerical results are presented to confirm the efficacy of the proposed scheme.