The Research Of Memory Predistortion Technique

As the communication technologies'development, the bandwidth of the systems becomes wider and wider and the PAPR gets much higher too, which will lead to a more severe nonlinear distortion and obvious memory effects of the HPA, resulting in higher requirements of HPA linearization techniques. Although the linearization technology for memoryless HPA is mature, the predistortion results are not as good when there are memory effects.The traditional 2-Dimensional LUT can make a better compensation for HPA with memory effects. However the second index of current LUT is usually computed by uniform quantization of the compression function, this will cause non-uniform distribution and some tables cannot be updated sufficiency.This paper analyses the compression function of the 2-Dimensional LUT. With the decomposition formula, it obeys (2Q,2) F-distribution where Q is the memory depth. And then the probability density function is deduced. The probability density function can be adopted to compute the second index. The simulation process of this paper takes into consideration only the latest signal with reference to the previous input, which means M equals 1. The LUT size is determined first, and followed by a predistortion result comparison between different methods in terms of amplitude and power, respectively.The simulation results show the higher size the LUT's first index has, the more excellent the predistortion effects are and the length of the second dimension is related to the memory depth. When the second index is computed by amplitude, the result is superior to what is computed by power. And we can get a higher efficiency compensation for the nonlinear distortion with the proposed method in this paper.