Compensation of nonlinear distortion of power amplifiers in mobile communications is an important requirement for improving power consumption performance while
maintaining efficiency, since mobile phone became an essential accessory for everyone nowadays. This problem demands a good power amplifier model, in order to develop an effective predistortion system.
Current researches are focused on modelling and predistortion of power amplifiers with memory, as well as memoryless ones. Different methods for modelling are used, as the Volterra series, polynomial models, look-up tables, the Hammerstein models, the Wiener models, and artificial intelligence systems. For predistortion feedback, feedforward and digital predistortion techniques are used. Among digital predistortion methods there are artificial intelligence systems, used in this thesis for linearization of power amplifier. This thesis presents developed robust method for modelling power amplifiers without memory effects and gives a comparison of proposed method with least squares method. Also, this research presents two novel techniques based on artificial intelligence systems for modelling and predistortion of highly nonlinear power amplifier with memory. The first approach is based on artificial neural networks, while the second one uses adaptive fuzzy logic systems. Forward and inverse models of power amplifier are created with both proposed methods. Superiority of artificial intelligence systems over partial least squares method is presented. Developed models are employed in a cascade to make a linearized system. Verification of proposed methods is carried out through the signal performance parameters and spectra of measured signal and signal from predistortion system. The feasibility and performances of the proposed digital predistortions are examined by simulations and experiments. The comparison of proposed methods is given to present advantages/disadvantages of both methods. The achieved distortion suppression from 72.2% to 93.6% and spectral regrowth improvement from 11.4 dB to 16.2 dB prove that the proposed methods have great ability to compensate the nonlinear distortion in power amplifier.