Research On Linearization Techniques For Wideband Communication Systems

The improvement of channel capacity of wireless communication systems requires importing novel techniques,such as multi-input-multi-output(MIMO)systems,highorder modulation techniques,and concurrent multi-band transmitters.However,introducing these techniques brings new challenges to the linearization system.The MIMO systems need to integrate great amounts of Radio Frequency(RF)chains and antennas.Further increasing the number of integrated antennas usually needs to employ the hybrid beamforming architecture.In a hybrid beamforming architecture,one RF chain has to drive several power amplifiers(PA).This demands the digital predistortion(DPD)system to linearize these PAs at the same time.However,one DPD is hard to linearize several PAs due to the uniformity problem of the nonlinear behavior.High-order modulation technologies make a high peak to average power ratio(PAPR)of signals,which forces the transmitters to work in a high back-off region.To enhance the transmitters' power efficiency,commercial base stations mostly employ Doherty amplifiers,which have high back-off power efficiency.However,Doherty amplifiers always show complicated nonlinear behaviors due to the unique electronic topology,which needs special linearization techniques.Generally,the spectrum resource is dominated and distributed by the International Telecommunication Union(ITU)and each country's government.Faced with the shortage of spectrum resource,current multi-band transmitters have become one of the most used solutions for wireless service providers.However,the current multi-band transmitters produce intermodulation distortions due to the nonlinear effect of PAs.The frequencies of these intermodulation distortions depend on the center frequencies of carrier signals.As a result,it needs linearization techniques to compensate for these intermodulation distortions when the intermodulation distortions are inevitably close to the carrier signals.To solve the above problems,this dissertation researches the linearization techniques for wide-band communication systems from these three aspects.The innovations of this dissertation are as follows:1.To solve the one-to-multi linearization problem in hybrid beamforming MIMO systems,an averaged group linearization is proposed.Analog predistorters(APD)are applied to tune the PAs' nonlinear uniformity.The combination of APD and one shared digital predistortion(DPD)can linearize several PAs concurrently.As a metric of PA's nonlinear uniformity crossed normalized mean square error(CNMSE)is proposed.The best state of PA uniformity can be determined by searching the CNMSE of the PAs.In a one-to-two PA linearization experiment,the proposed linearization method can improve the adjacent channel power ratio(ACPR)20.5-22.0 dB,compared with traditional DPD enhanced by 7.1-8.8 dB.To overcome the disadvantages of time-consuming CNMSE searching and the inability to trace PA working conditions' variation,a step uniformization group DPD(SU-GDPD)is proposed.With the same linearization performance,the SUGDPD reduces the uniformity tuning time from exponential growth to linear growth with the number of PAs.Furthermore,SU-GDPD can continue to update the linearizer with the variation of PA nonlinearity.2.The high-order modulation technique expands the channel capacity of communication systems.However,it makes the PAs work at 5-15 dB back-off in 80%time.This makes the commercial base stations mainly apply Doherty amplifiers,which introduce nonlinear delay difference.In this dissertation,a memory crossed Volterra model(MCVM)is proposed based on the nonlinear delay of a two-branch system to model and linearize Doherty PAs with nonlinear delay difference.The model coefficient setting method and pruning ability are also discussed.The MCVM shows a 10 dB improvement in normalized mean square error(NMSE)compared with the generalized memory polynomial(GMP)model in a simulation to model a two-branch nonlinear system with delay difference.In linearization experiment with Doherty PA,Long Term Evolution(LTE)signal with 20 MHz bandwidth and 7 dB PAPR is applied.The proposed MCVM improves the ACPR by 21 dB,comparing with the GMP model enhanced by 1.8-2.5 dB.3.To solve the linearization problem when the intermodulation distortions are close to the carrier signals in multi-band transmitters,this dissertation proposes a multi-cell linearization structure and a combined linearization structure based on the generation mechanism of the intermodulation distortions.The corresponding nonlinear models for each intermodulation distortions are firstly derived.Considering the mutual interference of predistortion signals at different frequencies,modified direct learning linearization with multi-cell structure is proposed.Both simulation and experimental results show that the proposed multi-cell linearizer can efficiently mitigate the intermodulation distortions near the carrier signals.In a dual-band concurrent linearization experiment,the ACPR with intermodulation distortion can be improved by 16.1-16.6 dB.To compensate for the intermodulation distortion when the intermodulation distortions are overlapped with carrier signals,a combined linearization structure is proposed.By shifting the digital frequency conversion process to the nonlinear basis function generation step,combined models with intermodulation distortion are derived.In a concurrent dual-band linearization experiment with overlapped intermodulation distortions,the carrier frequencies are 0.7 GHz and 1.375 GHz.The lower adjacent channel of carrier one where the second order beat distortion appears shows 21.9 dB improvement in ACPR after the combined linearization,4.2 dB better than multi-cell linearization.The upper adjacent channel of carrier two,where the second harmonic distortion appears,shows 22.7 dB improvement after the combined linearization,1.9 dB better than multi-cell linearization.