非線性衛星CDMA接收器之設計

Abstract

展頻技術因具有抗干擾、多重進接、抑制複徑干擾、低攔截率等等特性，近年來逐漸廣為商用通訊界所使用。其中直接序列分碼多工（DS / CDMA）系統的容量因較分類（FDMA）與分時（TDMA）多工系統為大，在商用的通訊系統上漸漸佔有一席之地。雖然分碼多工系統的寬頻干擾（即同頻系統用戶所造成的干擾）之消除曾有不少人探討過，但這些研究皆未考慮有意干擾（jamming）的問題；而BPSK或QPSK信號對非線性失真的抵抗能力有限，也是一般軍事衛星系統不採用直接序列展頻信號的主因。 一般在衛星轉頻器中的高功率放大器多具有非線性的特性，在商業用途中為使系統維持良好運作，通常會選擇適當的輸入退據（input back - off, IBO）以達到最佳操作點，但是若想要得到較大的輸出功率，或是基於反干擾的考量，我們便得操作在飽和區（否則敵人很容易加入相位或振幅干擾）。由於這個因素所導致的強烈非線性效應，加上一般的衛星轉頻器並無處理信號（on - board processing）的能力，將使得反干擾多用戶接收器的設計更加麻煩。 在本論文中，我們將首先討論高功率放大器的非線性特性。其次，我們將使用Volterra模型對整個非線性衛星通道做識別（identification）的工作，並說明如何利用識別的結果來設計接收器。接著，我們會討論一種應用在非線性通道上的非線性等化器，並且模擬在有訓練用序列（training sequence）的情況下，非線性等化器的性能。為了同時解決寬頻與窄頻有意干擾的問題，我們提出了一種將最小輸出能量（minimum output energy, MOE）和非線性等化器結合在一起的新接收器。最後，我們並討論了合併解碼與等化的效果及其應用。

Recently, code division multiple access (CDMA) has become a popular air interface scheme for land mobile communication systems. Several proposed low earth orbit (LEO) satellite systems also select CDMA as their multiple access method. Because of the nonlinear characteristic of the high power amplifier (HPA) used in a satellite transponder, an appropriate input back - off (IBO) is often needed to guaranttee that the HPA operates in teh linear region. This is especially important for a direct - sequence / spread spectrum (DS / SS) CDMA system, for phase - modulated signals are very susceptable to nonlinear distortions. This thesis considers a special satellite CDMA system in which the transponder has very limited on - board processing capability. Furthermore, due to an anti - jamdesign consideration, the satellite's HPA has to operate in the saturation region. We first present two nonlinear equalization methods that alleviate the nonlinear distortion induced by the transponder's HPA in a single access environment. A receiver structure that consists of a Volterra nonlinear equalizer and an adaptive minimum - output - energy (MOE) decorrelator is then proposed to eliminate nonlinear distortion and suppress both multiple access interference (MAI) and narrowband interference (NBI). We also investigate the effect and advantage of combining equalization and decoding.