Research Of Concurrent Dual-band Power Amplifier Linearization

With the development of communication technology,the bandwidth requirement of transmission signals is increasing,and the spectrum resources are becoming more tense,Carrier Aggregation technology is gradually emerging.In order to ensure the quality of signal transmission and reliability of information transmission,the linearization of concurrent dual-band power amplifier is particularly important.This thesis mainly studies the digital pre-distortion of the concurrent dual-band power amplifier linearization and explores the linearization of the extended dual-input Doherty power amplifier as a concurrent dual-band power amplifier.The main work and innovations of this thesis are as follows:1.This thesis starts from the different architectures of concurrent dual-band transmitters,it studies the nonlinear distortion components of concurrent dual-band power amplifiers: in-band inter-modulation,in-band cross-modulation,and filtered out-of-band intermodulation.These distortion components derive four concurrent dual-band power amplifier behavior models for different scenarios from the baseband signal.2.In this thesis,through the analysis and comparison of the existing dual-band indirect learning architecture and the dual-band direct learning architecture,combining the advantages and disadvantages of above ones,an iterative learning algorithm is introduced for the linearization of concurrent dual-band power amplifier.Then this thesis proposes a novel dual-band digital pre-distortion architecture based on Iterative Learning Control.Experiment shows that under the condition of high gain compression of the wideband class-F power amplifier which on saturation output,the 20MHz-LTE signal of the two bands can be improved by 20 dB in ACPR and the NMSE improved by 25 dB after 5 iterations of the Gain-Algorithm.The Constant-Algorithm can achieve the expected target after iterated 8 times.These two algorithms have superior convergence.Under the same behavior model,the linearization performance of the proposed architecture is better than the dual-band indirect learning architecture,which is similar to the dual-band direct learning architecture: The ACPR is improved by 16.2dB for low band and 18.9dB for up band,NMSE improvement is 18 dB for low band and 18.4dB for up band.The iterative method is simple and the linearization performance is excellent.3.This thesis proposes a new dual-input Doherty signal separation architecture based on Iterative Learning Control.The process of feature extraction of dual-input Doherty and its nonlinear distortion are introduced.It is different from the traditional look as static functions single-input single-output digital pre-distortion nonlinear correction method,this thesis uses static peak path function to control the power amplifier efficiency and iterates carrier separation function to control the linearity of power amplifier,taking advantage of the additional freedom of the additional RF input to the dual-input Doherty amplifier,which allows the digital pre-distorter to compensate for residual error by identifying iterations.The experimental results show that under the specified signal,the iterative algorithm can improve the ACPR of 20MHz-LTE signal by 30 dB,the NMSE improves 32 dB with the efficiency only reduced by 2%.The carrier separation function is fitted by the Self-Intermodulation Phase-Align model,the ACPR of other different signals can better than the communication standard-45 dBc,which verifies the superiority of linear and efficiency control of the dual input Doherty signal separation architecture based on iterative learning control.