Estimating Nonlinear Parameters Present in OFDM-based System Using Non-linear Least Squares

Abstract

Wireless communication systems are omnipresent in our everyday life and high standards regarding capacity, re- liability, speed and power are naturally expected. In order to satisfy these requirements, more sophisticated and efficient technology is needed. This is especially true for the integrated amplifiers at the transmitter side, which are responsible for the majority of the power consumption. In order to obtain high gain at the low power levels, these amplifiers are operated into their nonlinear region. However, many communication schemes cannot handle the nonlinear distortion generated by such amplifiers. The nonlinear distortions generated at the transmitter side can be compensated through the use of digital predistortion. This method compensates the non-linear effects by first esti- mating a non-linear model for the amplifier, and then applying the inverse model onto the data during transmission such that the nonlinearities are canceled. The main advantage of this approach is that it allows to use the classical linear communication scheme once the precompensation is performed.

In this work, Special attention goes to the estimation of the nonlinear model used for the predistortion. Most nonlinear models are not only nonlinear in behavior but also nonlinear in their parametrization, which requires spe- cial care during estimation of the model parameters. Performance of the simple linear least square estimator and nonlinear least square estimator is investigated using a National Instrument student setup.