Compensating method for nonlinear distortion by using neural network in digital satellite broadcasting

摘要

The transmission path in digital satellite broadcasting is characterized by a linear frequency dependence on the amplitude-frequency characteristics, group-delay-frequency characteristics, and AM/AM and AM/PM conversion effects due to traveling-wave tube amplifier (TWTA) nonlinearity. The AM/AM conversion effects result from the nonlinearity between the input and output amplitudes, while the AM/PM conversion effects result from the nonlinearity between the input amplitude and the output phase of the satellite transponder. We have developed a method for identifying and compensating for the linear frequency dependence and nonlinearity of a TWTA. This method combines an equalizer and a linearizer through the use of an adaptive algorithm. The linearizer is based on the neural-network concept. In our method, the reversal characteristics of the TWTA nonlinearity are identified, and the TWTA nonlinearity is compensated by using an error back-propagation algorithm based on a neural network. In this paper, we describe the development, features, and performance of our hybrid LMS-neural-network distortion-compensating method.