透過您的圖書館登入
IP:3.129.59.176
  • 學位論文

應用於直序分碼多工下鏈系統的低複雜度新型接收機設計

New Reduced-Complexity Receiver Designs for DS-CDMA Downlink Systems

指導教授 : 李學智

摘要


近年來,直序分碼多工已成為最廣為使用的行動通訊系統之一,隨著使用者對於數據通訊需求的增加,勢必得使用更大的頻寬進行無線傳輸,因此多路徑通道的效應變得更為顯著。無線通訊中的其中一項挑戰即為對抗多路徑通道所產生的影響。 在直序分碼多工下鏈系統中,廣義耙狀接收機能夠非常有效的抵抗由多路徑通道所引入的干擾,但廣義耙狀接收機必須求解一個大維度的反矩陣。在本篇論文中,我們提出一個使用最大比率結合-最小方均誤差的新型耙狀接收機。我們提出的新型耙狀接收機是傳統耙狀接收機和廣義耙狀接收機的一個折衷。此新型耙狀接收機中的訊號耙指以最大比率進行結合,並且擺放額外的雜訊耙指用來估計訊號耙指當中的雜訊和干擾成分並加以消除。我們也提出了一個高效率的耙指選擇演算法,期望夠有效率地找出有用的雜訊耙指。此外,我們將各種耙狀接收機推廣到接收端分集技術以及多輸入輸出的系統當中。模擬結果顯示此新型耙狀接收機有可能只需要較低的複雜度便能夠超越廣義耙狀接收機的效能,並且證明高效率耙指選擇演算法確實能夠選擇出有用的耙指。 另一方面,我們也針對最小方均誤差等化器當中的反矩陣問題進行討論。在本篇論文中,我們提出利用紐曼級數近似法和量子力學當中的雷利-薛丁格微擾理論,以修正快速傅利葉轉換近似法所求解出的近似反矩陣。此外,我們利用離散希伯特轉換導出一個求雷利-薛丁格微擾修正項的快速演算法。藉由分析訊號-干擾及雜訊比的分布情形,我們發現雷利-薛丁格微擾理論可以有效的降低訊號遭受嚴重衰減的機率,並且進一步提出兩種方法使得訊號-干擾及雜訊比有更理想的統計特性。對於一個最小方均誤差等化器,模擬的結果對於顯示紐曼級數微擾法無法有效修正快速傅利葉轉換近似法的結果,但雷利-薛丁格微擾法能夠極為有效的改善快速傅利葉轉換近似法的數值誤差,並且以極低的複雜度逼近理論的最佳效能。

並列摘要


In recent years, DS-CDMA systems have gained its popularity and become the most widely used mobile communication systems. However, as the transmission bandwidth increases, one of the primary challenges in wireless communication is to overcome the effect of multipath channel. The G-RAKE receiver is capable to mitigate the interference that is introduced by the multipath channel. The main defect of a G-RAKE is that one needs to evaluate a large matrix inversion. In this thesis, we propose a new MRC-MMSE-RAKE receiver which is a compromise between conventional RAKE and G-RAKE receivers. The signal fingers are placed to collect signal energy and combined with MRC. Additional noise fingers are introduced in order to suppress the interference of the MRC output by the MMSE interference estimation. We also propose an efficient finger selection algorithm (EFSA) to select valuable noise fingers. In addition, we extend the RAKE receivers to the receive diversity scheme and the MIMO transmission. From the simulation results, the MRC-MMSE-RAKE may outperform the G-RAKE with lower complexity, and the proposed EFSA can be utilized to select noise fingers effectively. On the other hand, we also design two types of MMSE equalizers for DS-CDMA downlink systems. We propose using the Neumann series perturbation (NSP) method and the Rayleigh-Schrödinger perturbation (RSP) method to correct the FFT-based approximation in matrix inversion. Besides, an algorithm for finding the RSP correction is also derived. By analyzing the distribution of SINR, it is found that the RSP is able to reduce the probability of the occurrence of deep fading events, whereas the NSP can not be. Several strategies were proposed to enhance the statistical characteristic of SINR in RSP. From the simulation results, the RSP is able to correct the inaccuracy of the FFT-based approximation and it may achieve the performance of the optimal MMSE equalizer.

參考文獻


[1] R. Price and P. E. Green, Jr., “A Communication Technique for Multipath Channels,” Proc. IRE 46, pp. 555-570, 1958.
[2] G. E. Bottomley, T. Ottosson and Y.-P. E. Wang, “A Generalized RAKE Receiver for Interference Suppression,” IEEE J. Select. Areas Commun., SAC-18: 1536–45, Aug. 2000.
[4] J. Zhang, T. Bhatt and G. Klandyam, “Efficient Linear Equalization for High Data Rate Downlink CDMA Signaling,” in Proc. Conference Record of ASILOMAR, Volume 1, Nov. 2003, pp. 141-145.
[6] C. Berrou, A. Glavieux and P. Thitimajshima, “Near Shannon Limit Error- Correcting Coding and Decoding: Turbo Codes,” Proc. IEEE Intl. Conf. Commun. (ICC 93), pp.1064-70, Geneva, Switzerland, May 1993.
[7] C. Berrou and A. Glavieux , “Near Optimal Error Correcting Coding and Decoding: Turbo Codes,” IEEE Trans. Commum. , Com-44: 1261-71, October 1996.

延伸閱讀