New adaptive polynomial and neural network predistortion techniques for satellite transmissions

In order to allow maximum exploitation of the frequency spectrum, the current trend is to increase the spectrum efficiency by using high modulation techniques e.g. multilevel quadrature amplitude modulation (QAM). The use of High Power Amplifiers (HPA), e.g., TWT, in satellite communications channels causes severe nonlinear distortions to the transmitted signal. The transmitted signal is more sensitive to these nonlinear distortions when the number of symbols in the signal constellation increases. The effect of the HPA can be sufficiently reduced by backing-off the output signal levels from the amplifier's saturation point. However, this reduces the transmitted signal power and consequently the power efficiency. Therefore, there is a tradeoff between power efficiency and spectral efficiency. Nonlinear distortions of the HPA give rise to some unwanted effects such as, warping of the signal constellation. Also, nonlinear characteristics of the TWT cause an increase of the Adjacent Channel Interference (ACI) due to widening of the transmitted signal spectrum