[Zurück]

R. Dallinger, M. Rupp:
"Adaptive Digital Pre-distortion based on Two-block Models";
Vortrag: IEEE Forum on Signal Processing for Radio Frequency Systems, Vienna (eingeladen); 28.03.2011.

We summarise our recent research results about low complexity adaptive digital pre-distortion schemes for power amplifiers (PAs). The pre-distorter contains a memoryless nonlinearity followed by a finite impulse response filter (i.e., a Hammerstein model). The model used for the identification of the PA consists of the same blocks but with reversed order (i.e., a Wiener model). For both models, the nonlinearity is parametrised using orthogonal polynomials. The identification of the pre-distorter and the PA model is done by gradient type algorithms. In a first step, we compare the performance of different orthogonal polynomial bases with respect to the identification of a memoryless nonlinearity. Based on simulations and measurements, we evaluate the behaviour of Hammerstein pre-distorters adapted by either the indirect learning or the nonlinear filtered-x least mean squares algorithm. Finally, a robustness analysis based on the singular value decomposition allows to conjecture the robustness of Wiener models adapted by gradient type algorithms.

Projektleitung Markus Rupp:
Signal and Information Processing in Science and Engineering - Entwicklungsmethodik