Combating Nonlinear Power Amplifier Effects in Multicarrier Systems

Multicarrier modulation is widely accepted as a powerful technology to satisfy the increasing demand of high data rate transmission. The inherent large envelope fluctuations of multicarrier signals necessitate the use of power amplifiers (PA) with large dynamic range. However, the incompatibility between linear amplification and power efficiency poses challenges in deployment of practical wireline or wireless systems. When high power efficiency is mandatory, nonlinear power amplifier effects are inevitable and thus system performance is degraded. In this dissertation, several novel methods based on signal design and processing techniques are proposed to combat nonlinear power amplifier effects in multicarrier systems, specifically focusing on boosting power efficiency and improving bit error rate in orthogonal frequency division multiplexing (OFDM) and multiple access OFDM (OFDMA). First, with emphasis on easy implementation and low cost at the transmitter, two novel peak windowing schemes, with asymmetric window functions and simple coefficient optimization respectively, are proposed to handle the case of successive peaks in transmitted signals, which outperform existing methods. In addition, a novel receiver-oriented method is proposed to tackle the intrinsic challenge in OFDMA reception where an individual user has insufficient information of other users’ modulation to apply conventional decision-aided schemes. The proposed method includes two steps: improved peak localization and magnitude estimation with frame based alternating projection which does not require modulation information of the entire OFDM symbol, and has significantly enhanced performance compared with existing schemes, especially in the high signal-to-noise regime. Furthermore, we propose a joint design of transmitter-oriented scheme of tone reservation with clipping and receiver-oriented scheme of frame-based alternating projection which opens up a novel way to deal with the nonlinear power amplifier effects. The tight requirement of sparsity level in clipping noise imposed by compressed sensing framework is relaxed with the novel formulation based on frame theory and projection over convex sets and thus significant bit error rate performance improvement is achieved. Moreover, a hybrid scheme that combines two existing single-antenna schemes, namely erasure pattern selection and Fourier projection algorithm, is designed to address the issue for the multi-antenna case and simulations demonstrate it outperforms a popular existing scheme.