Study On Digital Predistortion Technique For Power Amplifier With Memory Effects

Power amplifiers (PAs) are critical elements of a transmitter in modern wireless communication systems. In order to increase power efficiency, PAs are often driven into the nonlinear region near the saturation point. For compensating the nonlinear distortion of a power amplifier (PA), linearization techniques are studied widely, and predistortion technique is one of the promising linearization techniques. Nowdays, predistortion techniques, which take emphasis on overcoming the memory effects of PAs, are the research hotpoint.This thesis studies the predistortion techniques based on Hammerstein model, and is focused on the realization form of linear subsystem in Hammerstein predistorter and the predistortion structure for identifying the predistorter parameters with the high-efficient lease square algorithm. Specifically, this thesis completes the following tasks.(1) Propose an improve lookup table method. In a Hammerstein predistorter, the lookup table is often used as the realization form of the nonlinear subsystem. In the lookup table method, the quantization error is generally larger in the region of small signals. To solve this problem, considering the merits of the uniform quantization method and the characteristics feature of strong nonlinearity when a PA operates in the region of large signals, uniform quantization used in different subsections is adopted. The simulation results show the efficiency of the proposed method.(2) The method that partial update least mean square (PU-LMS) algorithm is used to identify the nonlinear subsystem of a PA is given. For identifying the linear subsystem of the traditional Hammerstein predistorter, there are more parameters that need to update. To solve the problem, PU-LMS algorithm is adopted to update the parameters of the linear subsystem. That is to say, the strategy that the parameters are divide into groups to update is adopted, throuth which the number of parameters is decreasea in iteration so as to decrease the calculated amount of the identification algorithm.(3) Propose a novel Hammerstien predistorter. The memory effects of PAs is actually a delayed response of apparatus that cannot follow the fast change of signals, which is different when the input power changes. Considering these characteristics of memory effects, a two-dimensional filter lookup table is adopted as the linear subsystem of a Hammerstein predistorter. In the predistorter, the second-dimensional index is defined to reflect the change rate of signal envelope for better describing the inverse memory effects of PAs efficiently. The theoretical analysis and simulation results demonstrate that the proposed Hammerstein predistorter can more efficiently improve the linearization performance for the PAs with memory effects.(4) Propose a novel predistortion structure for identifying the parameters of a Hammerstein predistorter by using high-efficient least square algorithm directly. When high-efficient least square algorithm is used to identify the parameters of a Hammerstein predistorter, there are two problems:one is the high complexity of the algorithm; and the other is the interference between the two subsystems. To solve these problems, a novel structure is proposed. By formula derivation, it is proved that the extracted error signals from the proposed structure are equal to ideal error signals of subsystems in predistorter. Consequently, the parameters of Hammerstein predistorter can be identified by high-efficient least square algorithm directly. The theoretical analysis and simulation results demonstrate that the Hammerstein predistorter based on this proposed structure can efficiently compensate the nonlinear distortion of PAs with memory effects.