Novel DLA-based digital pre-distortion technique for power amplifier

As a coefficients extraction method of power amplifier linearization technique based on the direct learning architecture (DLA), the traditional iterative method has the merits of less iteration times and high convergence rate and is commonly used. Its computation, however, because of the two order partial derivatives of the cost function is quite complex. In order to overcome the shortcomings of the traditional iterative method, a new DLA-based pre-distortion technique, which combines steepest descent with instantaneous gain, is proposed in this paper. By employing steepest descent method, the coefficient update method for digital pre-distorter avoids complicated two order partial derivatives of traditional iterative methods. Furthermore, the concept of instantaneous gain of power amplifier is introduced to avoid solving power amplifier model during the iteration. Simulation results show that the proposed technique can improve the convergence speed and performance of the algorithm.