Adaptive Linearization of Power Amplifiers in Digital Radio Systems

Progress in high-speed data transmission over radio channels has lagged behind that of the voiceband channel. Inherent difficulties associated with implementing automatic equalizers are partially responsible, but the application of multilevel quadrature amplitude modulation (QAM) also has been inhibited by the amplitude (AM/AM) and phase (AM/PM) nonlinearities present in radio-frequency (RF) power amplifiers. Recent work1 has evolved design principles showing a possibility of substantial improvements in QAM performance over linear fading radio channels. A crucial obstacle to achieving these gains is the nonlinear distortion introduced by power amplifiers. Attempts to re1019 alize high-speed data transmission over these channels force consideration of methods to cope with this nonlinear distortion. One approach is to back off from saturation sufficiently so that the signal level is restricted to the linear range of amplification. The required amount of power back-off can be several decibels, resulting in an inefficient operation of the power amplifier. Moreover, the achievement of a given desired level of average transmitted power would require the use of a large, expensive, and high-power-consuming amplifier. It has been realized2"7 that some improvements in this regard can be obtained by using fixed signal predistortion circuits prior to amplification. Such circuits, however, cannot compensate for drifts in power amplifier nonlinearities caused by temperature changes, dc power variations, and component aging.