Power Efficiency Enhancement and Linearization Techniques for Power Amplifiers in Wireless Communications

Wireless communication systems require Power Amplifiers (PAs) for signal transmissions. The trade-off between power efficiency and nonlinear distortion in PAs degrades the communication performance. Thus, power efficiency and nonlinearity are two main concerns of operating PAs in communication systems. Nonlinear behavioral models are typically used to quantify and mitigate the distortion effects of PAs on communication systems. This dissertation presents an estimation approach for modeling and linearizing the PA Amplitude-to-Amplitude (AM/AM) nonlinearity using the design specifications of PAs, such as gain, the third-order intercept point, and 1dB compression point. Furthermore, an enhanced approach for modeling solid-state power amplifiers is developed by modifying the Saleh empirical model. The Envelope Tracking (ET) technique for PAs has been a popular power efficiency enhancement in modern cellular systems. However, the time-varying effects of the supply voltage impacts the PA linearity. Therefore, an accurate behavioral model for PA with ET has become an important research effort to characterize the effect of dynamic supply voltage on both the amplitude and phase nonlinearities. Furthermore, the empirical models of ET PAs are widely used to improve PAs linearity by using Digital Predistortion (DPD). This dissertation develops an extended modeling approach to characterize the AM/AM and Amplitude-to-Phase (AM/PM) conversions as well as account for the impact of the time-varying supply voltage on the ET PAs. Memory effects, due to energy storage elements (e.g. capacitors and inductors) in ET PA circuits in addition to the temperature variation of integrated circuit, are modeled using digital filters (finite impulse-response filters) in series with the static AM/AM and static AM/PM nonlinearities. A least-squares approach is mathematically derived for estimating the model coefficients of ET PAs. The model identification of many coefficients requires high computational cost in Float Point Operations (FLOPS), such as multipliers and adders. In addition, the computational cost in FLOPs of a complex number is equivalent to (2-6) times the cost of real numbers. The estimation complexity of the ET PAs model in this work requires around half the number of FLOPS compared to the state-of-the-art behavioral models. This is because the modeling approach in this work consists of real coefficients and a lower number of model parameters. A DPD model is derived in this dissertation to compensate for both the AM/AM and AM/PM nonlinear distortions in ET PAs. A dual-input single-output function architecture is calculated for the DPD model to compensate for the nonlinearities in the AM/AM and AM/PM conversions contributed by the time-varying supply voltage in the ET system. Both the proposed AM/AM and AM/PM DPD models exhibit lower numbers of coefficients, which result in reduction of the identification complexity compared to the state-of-the-art DPD models. The proposed behavioral models of the ET PA and DPD are both evaluated in the time and frequency domains, as well as compared to the state-of-the-art models in terms of model accuracy and estimation complexity