A nonlinear equalization technique for digital communications

The problem addressed in this dissertation is correction of signal distortions caused by bandwidth-power efficient modulation schemes in satellite communication channels. Communication satellites make use of high power amplifiers (HPA) that have nonlinear input-output characteristics. These high power amplifiers are employed in the nonlinear region of their operation so that maximum power can be derived from them. If these high power amplifiers process a variable-amplitude signal while they operate in the nonlinear region, the output signal will have distortions that are nonlinear in nature and cannot be corrected with traditional linear adaptive filtering techniques. To avoid an increase in symbol errors due to nonlinear distortions of variable amplitude signals, these high power amplifiers operate in the linear region of their characteristics; that is, they operate at less than their maximum output power. This phenomenon is known as power back off. Communication systems are, therefore, sacrificing precious power to use bandwidth efficient modulation schemes.; In this dissertation, we employ a Volterra-series-based nonlinear adaptive filtering technique, to counter the nonlinear distortions introduced by the high power amplifiers present in the satellite communication channel, when variable amplitude signals are transmitted through these communication channels, thus making it possible to use a bandwidth-power efficient communication scheme. We considered quadrature amplitude modulation (QAM) technique in our system. We analyzed a communication system that includes both a linear passband filter and an amplifier in the nonlinear region of operation, thereby avoiding any underuse of power. First, we develop a mathematical model of the communication system which makes use of an analytical approximation of the traveling wave tube (TWT) performance. Then we introduce a nonlinear equalization scheme and optimize its parameters using computer simulations. The optimized equalization scheme is then applied to all possible combinations of three overlapping consecutive symbols in order to verify the quality of restoration of the original signal. Then we consider the effects of additive noise and determine the symbol error rate/bit error rate in the system. The results are compared with those for linear adaptive filters for different values of signal-to-noise (S/N) ratio. It turns out that the nonlinear equalization counters the effects of both intersymbol interference as well as the nonlinear distortion introduced by nonlinear elements in the system. Our simulation results show that nonlinear adaptive filter technique performs significantly better than linear adaptive filter technique.