| Based on an overview of the development of the next generation wireless communication, a promising wireless digital modulation technique called OFDM is studied in this thesis. OFDM,a high rate transmission with many outstanding advantages, has one disadvantage: it has a high PAPR which makes it very sensitve to the nonlinearities of amplifier. To mitigate the nonlinearity distortions to OFDM signal and improve PA's efficiency, it is necessary to compensate the nonlinearities introduced by the amplifier. For this purpose, a lot of linearization methods have been investigated including back-off, feed-forward, feed-back, LINC and predistortion. This paper mainly researches on the method of predistortion. Firstly, an application of recurrent neural network to adaptive predistorion is briefly introduced here, then algorithms of adaptive predistortion based on LUT and polynomial are discussed in detail. Also a modified algorithm is proposed, and computer simulation results show that it does improve further the predistortion performace and evidently compensates the nonlinearities of memoryless HPA. Meanwhile, predistortion based on polynomial with memory is applied to the HPA with memory and is proved of its validity by simulation. |
PDF Full Text Request