無合成器之 LINC-OFDM系統最大似然偵測

Abstract

眾所周知，OFDM(Orthogonal Frequency Division Multiplexing; OFDM)系統的傳送訊號有很高的PAPR(peak-to-average power ratio; PAPR)值，因此其功率放大器(power amplifier; PA)需有較大的線性放大區使系統能正常操作，這造成PA為射頻(radiofrequency; RF)電路中最消耗功率的元件。LINC(Linear amplification with nonlinear component; LINC)技術為降低高PAPR系統功率消耗的方法之一，LINC系統可將輸入訊號利用效率較好的非線性放大器進行線性放大。然而，在LINC系統中有一關鍵的元件即功率合成器(power combiner)，其設計及實現有最高的難度，為了避免使用合成器，有研究者提出了無合成器之LINC系統，但在非理想的通道條件下，無合成器之LINC系統效能很不理想。在本篇論文中，我們研究提出無合成器之LINC-OFDM系統，並提出一最大似然(maximum likelihood; ML)偵測演算法以改善其效能，使用我們的方法可以非理想的通道條件能有效地操作。模擬結果顯示，出我們所提出的演算法能有效的提升無合成器之LINC-OFDM的效能，在很多情況下甚至比傳統OFDM系統的效能表現更好。此外，我們也提出了數種降低ML運算複雜度的方法。

The transmit signal in Orthogonal Frequency Division Multiplexing (OFDM) systems is known to have high peak-to-average power ratio (PAPR). Due to this property, the power amplifier (PA) of the system must operate in a wide linear region, making it the most power-hungry device in the RF circuit. The linear-amplification-with-nonlinear-component (LINC) technique has been developed to reduce the power consumption in high-PAPR systems. By using the LINC architecture, nonlinear PAs with high power efficiency can be used to linearly amplify the input signal. However, a critical component in LINC transmitter, named power combiner, is difficult to design and implementation. To avoid the use of the combiner, a combinerless LINC system is later proposed. Unfortunately, the performance of the system is poor in some non-ideal channel conditions. In this thesis, we study the combinerless LINC-OFDM system and propose a new maximum likelihood (ML) detection algorithm to improve its performance. With the proposed algorithm, the LINC-OFDM system can be effectively operated in non-ideal channel environments. Simulations show the proposed algorithm can significantly enhance the performance of the combinerless LINC-OFDM systems. In most cases, the proposed algorithm can even outperform the conventional OFDM systems. We also propose several methods to reduce the computational complexity of the ML algorithm.