Crest Factor Reduction and Digital Predistortion of RF Power Amplifiers Driven by Carrier Aggregated Signals
MetadataShow full item record
Advanced modulation techniques and access technologies are enabling higher data-rate communication at the cost of stringent signal requirements affecting radio transceivers’ efficiency and cost. One of the major problems is the high peak to average power ratios (PAPRs) of modern signals, which are imposing contradictory linearity and efficiency constraints on radio frequency power amplifiers (PAs). One key solution to extending a PA’s linearity and efficiency is the application of digital predistortion (DPD) techniques. A second approach is the deployment of crest factor reduction (CFR) techniques to effectively reduce the PAPR of the signal. To date, significant progress has been reported in the literature in both of these areas, but only in the context of single-band multi-carrier signals. Recently, in an attempt to extend bandwidth and increase spectral efficiency, new standards have been adopting multi-band communication schemes. These signals, also known as carrier aggregated signals, introduce two new challenges to the design of efficient radio systems. First, the wide spectral separation of the different component carriers limits the conventional modeling schemes. Extending or revising classical single-input single-output formulations is not viable, and a multi-input multi-output (MIMO) methodology needs to be developed. A second challenge is the projected deployment of up to five component carriers per signal. This implies the expensive deployment of up to five predistorters, five transmitter observation receivers (TORs) and five training engines (TEs). This thesis presents a number of contributions addressing the above challenges and paving the way for the deployment of carrier aggregated radios. The first contribution of this thesis is the development of MIMO CFR and dual-band DPD modules to enable the low-speed baseband processing of carrier aggregated signals. The low-speed CFR solution is extended to cover different deployment scenarios. Next, the conventional PA behavioral modeling approach is reviewed and reformulated to effectively design a novel dual-band DPD scheme. The second contribution is the efficient hardware implementation of predistortion algorithms. A symbolic optimization approach is proposed to enable the joint optimization of the two dual-band predistorters. Next, a time-shared 1-TOR 1-TE real-time adaptive learning approach is proposed to efficiently linearize a dual-band PA.