應用於低耗能正交分頻多工系統的LINC傳送機設計

Abstract

正交分頻多工（Orthogonal-frequency-division-multiplexing; OFDM）已廣為使用於現今的無線通訊系統中。然而OFDM的高峰值與平均功率比( peak-to-average power ratio; PAPR)，會造成功率放大器的效率低落。非線性元件之線性放大(linear-amplification- with-nonlinear-component; LINC)技術，可有效解決高PAPR的問題。在LINC架構下，原先的傳送信號會被分解成兩個固定振幅的成分信號，因此可使用高效率的非線性放大器來放大成分信號，最後使用功率合成器來結合兩個成分信號，達到高效能的線性放大效果。然而LINC技術存在許多實現上的問題，本論文旨在分析這些問題並且提出相對應的解決方案。首先，功率合成器雖然是LINC技術中重要的元件，卻有其實現上的困難，因此無合成器的LINC系統也被提出來解決此問題。然而，在真實環境中，無合成器的LINC系統的信號品質會被嚴重影響。儘管最大似然(Maximum-likelihood; ML)演算法可解決此問題，高計算複雜度使得ML演算法無法直接使用。在本論文的第一部分，我們提出了一種使用迴旋碼(Convolutional Codes)的編碼無合成器之LINC-OFDM系統。藉由使用條列式維特比演算法(List-Viterbi Algorithm; LVA)，可大幅減少ML演算法所需偵測的候選資料數量，因此可有效降低其計算複雜度。模擬結果顯示，我們提出的編碼無合成器之LINC-OFDM系統，可以與傳統OFDM系統有相同的效能，但卻有較低的消耗功率。 除了功率合成器，在LINC架構下，兩個成分信號的放大過程中也容易產生不匹配的問題，其中包含了放大倍率、相位、以及時間延遲上的不匹配。LINC系統內的不匹配會造成傳送信號的失真和頻譜洩漏效應。在本論文的第二部分，我們探討如何解決LINC傳送機的不匹配效應。這部分的論文分為兩個層面，首先我們分析不匹配效應所造成的信號失真。先前的研究者皆未考慮到接收機內等化器的效應，因此高估了失真的影響。而我們分析了使用LINC架構傳送的OFDM信號在線性等化後的誤差向量幅度( Error Vector Magnitude; EVM)，其結果可用於定義在系統可以容許的最大不匹配。接下來我們針對放大倍率、相位、以及時間延遲上的不匹配，提出了相對應的適應性補償演算法。模擬結果顯示，我們的分析結果很準確，而提出的補償演算法也比先前研究有著較好的效能。 當LINC傳送機使用於雙頻傳輸時，不同頻帶上的成分信號會產生非線性的效應。數位預失真技術，已被証明可有效地進行功率放大器的線性化。然而傳統的數位預失真技術皆是為單頻傳送機設計的。為解決此問題，近年有學者提出了基於二維多項式的雙頻數位預失真技術，然而高計算複雜度為此做法的最大缺點。一般來說，基於查表法的數位預失真技術，相較於多項式的做法，有著較低的複雜度，但未有研究探討如何利用查表法來進行雙頻傳輸的數位預失真。在本論文的第三部分，我們提出了一種基於二維查表法的Hammerstein預失真技術。為了計算進行預失真時所需的參數，我們針對梯度下降(gradient descent; GD)以及遞迴最小平方(recursive least-square; RLS) 演算法，提出了相對應的適應性參數訓練演算法。模擬和實驗結果証明了我們的做法其效能和現存的二維多項式數位預失真近似，但我們所提出的參數訓練演算法卻有較低的計算複雜度。

Orthogonal-frequency-division-multiplexing (OFDM) has been used in many modern wireless systems. However, it is well known that its peak-to-average power ratio (PAPR) is high, yielding the low power efficiency problem. The linear-amplification-with-nonlinear-component (LINC) transmitter decomposes the input signal into two constant-envelop component signals, amplifies each signal with a high-efficient nonlinear power amplifier, and then combines the resultant signal with a power combiner. It has been seen as a remedy for the high PAPR problem. However, there are several issues in the conventional LINC transmitter, precluding its wide applications. This dissertation aims to propose new methods solving these problems. First, the power combiner, a key component used to combine the amplified signals, is difficult to implement. A combinerless LINC system has been proposed to solve the problem. Unfortunately, the performance of combinerless LINC-OFDM systems can be seriously degraded in real-world environments. The maximum likelihood (ML) receiver can be used to solve the problem; however, its computational complexity is prohibitedly high. In the first part of this dissertation, we propose a coded combinerless LINC-OFDM system, including a convolutional encoder and a list Viterbi algorithm (LVA) decoder, to address the problem. The LVA can provide a small number of candidates for the ML detector, dramatically reducing the required computational complexity. Simulations show that the performance of the proposed combinerless LINC-OFDM system can outperform the conventional OFDM system. However, the power consumption is much lower.

Besides the combiner, another issue is the imbalance problem presented in the LINC transmitter. The two LINC paths may be imbalanced in either gain, phase, or timing-offset, causing both the in-band and out-of-band distortions. In the second part of this dissertation, we consider the imbalance compensation in the LINC transmitter. The contribution of this part is twofold. First, we analyze the in-band distortion caused by the imbalances. Most existing analyses do not consider the equalization conducted in the receiver, resulting in overestimated results. We incorporate the effect into consideration and derive a closed-form expression for the error-vector magnitude (EVM) of the OFDM signal. The analytical result can be used to evaluate the effect of the imbalances and specify the matching requirement for target signal quality. Second, we design an adaptive method, which can effectively compensate for the gain, phase, and timing-offset imbalances. Computer simulations show that our analysis is accurate and the proposed compensation method significantly outperforms the existing.

When the LINC transmitter is applied for the dual-band transmission, the combined component signals with different carrier frequencies will introduce nonlinear effect. Digital predistortion (DPD) has been proved to be an effective remedy for this problem. However, the general DPD technique for single-band transmitters cannot be directly extended to dual-band. To solve the problem, DPD with two dimensional memory polynomials was then proposed. The disadvantage of this approach is that its computational complexity is high. The look-up-table (LUT) based DPD, developed for single-band transmitters, can effectively solve the problem. Unfortunately, LUT-based DPD for the dual-band transmitters has not studied yet. In the third part of the dissertation, we propose a two-dimensional LUT-based Hammerstein DPD to solve the problem. For DPD training, we proposed new adaptive algorithms using gradient-descent and recursive-least-squares (RLS) methods. With the proposed method, the computational complexity can be significantly reduced. Finally, simulations and experiments are conducted to demonstrate that the performance of the proposed DPD is comparable to that of the polynomial-based DPD.