Research On Baseband Predistortion Of Power Amplifiers

In order to solve the contradiction between the scarce spectrum resource and the higher information transmitted rate, the modern communication systems commonly adopt the linear modulation methods with high spectrum efficiency. However, under the condition of the nonlinearity, the performance of these efficient modulation techniques will deteriorate badly. But in the actual applications for improving power efficiency, the power amplifier commonly works in the nonlinear area near the saturation. Then the amplifier is able to bring very severe nonlinearity distortion, which results in the spectrum expansion of the output signal and the signal distortion in the band, brings the adjacent-channel interference and the fall of the system performance. Hence, in order to consider both system performance and power efficiency, the linearization of power amplifier has been the hot spot in the wireless communication research.The main contents of the thesis are as followed:1. The nonlinear characteristic of power amplifier is analysed, then the model of memoryless power amplifier and several common linearity techniques are introduced.2. In the predistortion research based on the memoryless amplifier, the method of LUT is firstly studied, simulation results show LUT can compensate the nonlinearity effectively. And in the same size of LUT, the performance of the LUT with uniform quantization in subsection is better than the traditional LUT. Secondly, the method of polynomial is studied. Based on the LMS algorithm and the RLS algorithm, simulation results show it is an effective method of linearity. Finally, the iterative steps become alterable in the LMS algorithm of the polynomial, simulation results prove that the convergence speed and precision are better than the original LMS algorithm, the performance is closed to the RLS algorithm.3. In the predistortion research based on the memory amplifier, on the foundation of introducing several common models of the memory amplifier, the memory polynomial predistortion is studied, simulation results show that it can compensate the nonlinearity effectively based on the Wiener model and the Wiener-Hammerstein model, the constellation is the same as the source signal on the whole, and the power out the band falls 20dB and 26dB respectively.4. A new structural predistortion is proposed, the method is simple and easy to come true, but the power efficiency will fall a bit. Simulation results show this method can compensate the nonlinear distortion well based on the memory polynomial model, the power out the band falls 28dB, improves 13dB compared with the memory polynomial predistortion.