Low frequency feedforward and predistortion linearization of RF power amplifiers

Linearity is an area of great concerns in the development of communication systems. As the bandwidth of modulated signals becomes wider, RF power amplifiers (PA) have become a serious bottleneck due to their nonlinear response. This thesis gives a review of existing linearization systems commonly used in cellular phone basestations and which could be potentially applied to mobiles. Special focus is placed on predistortion linearization which is becoming prevalent. An alternative implementation of postdistortion often referred as low frequency feedforward linearization is also investigated. The existing issues associated with both low frequency feedforward and predistortion linearization are analyzed in details. Memory effects in PAs are identified as the origin of the degradation of the performance of linearization systems as the bandwidth is increased. Our research is concerned among other things with the development of improved RF predistortion algorithm accounting for such memory effects. In this thesis, we present theoretical work where the Volterra representation is used to analyze the performance of both predistortion and low frequency feedforward linearization systems in the presence of memory effects. A new real time algorithm is then developed for predistortion linearization which accounts for memory effects. This real time digital algorithm is then implemented in a FPGA testbed and verified using multisines and WCDMA modulated RF signals. This thesis concludes with a discussion of possible extensions of this work combining the various novel linearization schemes proposed.