Adaptive digital baseband predistortion

Abstract

The radio frequency (RF) power amplifier (PA) is the most power consuming element in a wireless transmission system, and can account for more than 50 percent of the total power consumed by the transmitter [1]. Improving the PA efficiency saves energy and drives down the overall system costs. A study described in [2, p. 13] provides a compelling reason for PA linearization: application of PA linearization technologies can yield annual savings of millions of dollars for a typical network service provider.Efficient PAs are usually nonlinear. Nonlinearity generates both in-band distortion and out-of-band interference, which manifest in terms of transmitter error vector magnitude (EVM) degradation and spectral regrowth. In wireless communication systems, many signal formats, such as Code Division Multiple Access (CDMA) and Orthogonal Frequency Division Multiplexing (OFDM) transmission, have been introduced to improve spectrum efficiency and data rate. However, these non-constant-envelope signals are not power efficient in the presence of nonlinear PAs as large back-offs are needed for linear transmission. In most commercial wireless communication systems, PAs are still the dominant source of signal quality degradation since the PAs are usually biased in a mildly nonlinear region to gain a reasonable amount of efficiency.Power amplifier linearization techniques can be sought to improve the linearity of the PA while still achieving a certain degree of power efficiency. Ideally, after linearization, the linear region of the PA is extended close to the saturation point of the PA, allowing a smaller power back-off than the original PA, which in turn improves the PA power efficiency. Power amplifier linearization has been widely studied. Available techniques include feedback, feedforward, and predistortion [2][3][4][5].Feedback is widely used in control systems. In the context of PA linearization, if we can extract the error signal as the difference between the (scaled) input and output to the PA, a negative feedback system can be constructed by feeding the error signal back to the PA input [6,7]. The gain-bandwidth trade-off and the instability problem, however, limit the feedback linearization performance of RF PAs, especially for wide-band applications.Feedforward linearization reproduces the error signal and compensates for it at the PA output [8]. In contrast, with feedback linearization, the compensation takes place at the Digital Front-End in Wireless Communications and Broadcasting, ed. Fa-