Investigation On Digital Linearization Technology For Short Wave Transmitter

Short-wave(SW)communication system is widely used in fields of emergency response,disaster relief,military communication and global broadcasting with the advantages of remote communication with no relay stations and high flexibility.Power amplifiers(PA)play an important role in the SW communication system.The nonlinear distortion which caused by PA contains not only the spectrum regeneration of fundamental adjacent channels,but also the harmonic interference that affects signal transmission in higher frequency bands.Digital predistortion(DPD)as one of the widely used linearization techniques,can effectively improve the fundamental distortion in shortwave transmitter.However,for harmonic cancellation(HC),the traditional way of using analog filter banks has many disadvantages such as high cost,and severe insertion loss.The demand for more effective digital harmonic elimination methods has increased.Focuses on the digital linearization technology in SW transmitter,this dissertation investigates simpler and more efficient digital HC models and architectures.Combining the DPD technology,the relatively complete digital linearization scheme for SW transmitter is ultimately formed ultimately by solving the mutual influence between between the DPD and HC.The main contents and innovations of this dissertation are listed as follows:1.The derivation of harmonic behavioral model characterized by the real domain is proposed.Based on the analysis of the distortion components of harmonic signals in SW PA,radio frequency harmonic behavior models that can simultaneously characterize the harmonic signals at multiple frequencies is derived.The proposed models can be able to avoid the problem of adding cost to cancel multiple harmonics during baseband HC process.And the method of model performance evaluation is established by the iterative learning control theory.The experimental results about the modeling performance of SW PA under different signal conditions show that,the normalized mean square error(NMSE)of the second and third harmonic models were-40.21 dB and-30.50 dB when the carrier frequency was 10 MHz and the signal bandwidth was 100 kHz.Because of the low dynamic range of the third harmonic,the modeling error is relatively large.However,for HC,the proposed model can achieve similar results with Iterative learning control.2.The HC architecture based on digital harmonic injection is proposed,and corresponding coefficient identification algorithm is derived in this dissertation.The goal of the proposed architecture is using a single feedback link to cancel multiple harmonics simultaneously by injecting the harmonic pre-compensation signals at the front of PA.Comparing with the traditional feedforward baseband HC structures,the problems of complex implementation and high resource consumption of are simplied during the multiple HC process.Based on the amplitude-frequency response and phase-frequency response of PA,the amplitude alignment and phase compensation are proposed to improve the convergence speed and iterative stability.The experiment about the HC performance and amplitude-frequency and phase-frequency response compensation methods of SW PA under different signal conditions show that,the proposed architecture can simultaneously cancel the second and the third harmonics below-70 dBc when the carrier frequency is 5 MHz and the signal bandwidth is 100 kHz.After amplitude alignment and phase compensation,the iterative convergence speed can be increased more than 40%.3.The HC based on digital multi-tone mixing(DMTM)method is proposed.The principle of this method is to mix multiple harmonic signals into a narrow band range for coefficient identification.It can avoide the high computational complexity in full band direct sampling.The boundary conditions of multi-tone local oscillator intervals without aliasing are summarized by the analysis of the bandwidth variation of harmonic distortion.The low complexity HC architecture is proposed,and the mixed harmonic model is derived based on the mixed harmonic characteristics.The experimental results about the DMTM HC performance of SW PA under different signal conditions show that,the proposed method has similar performance with the full band direct sampling method.When the carrier frequency is 5 MHz and the signal bandwidth is 10 kHz,the second and third harmonics can be cancelled below-69 dBc,and the complexity of coefficient identification is reduced by more than 96 % compared with full band sampling.4.In order to linearize the baseband distortion and the harmonic interference synchronously in SW transmitter.A dual domain joint linearization scheme with synchronized coefficient identification is proposed by cascading baseband DPD and radio frequency(RF)digital HC.The impaction of pre-distortion signal envelope changes on harmonics and the impact of harmonic pre-compensation signal injection on baseband envelope is analyzed first in this dissertation.In order to cover the shortage of architecture and model of traditional joint linearization,a cascaded dual domain joint linearization architecture is proposed,and the mutual influence of the two linearizations under different coefficient identification ways is further discussed in this architecture.To avoid the above issues,a ceofficient identification method with synchronous update is proposed based on dual domain joint linearization,and the corresponding joint model is derived.The experimental results about the joint linearization performance of SW PA under the same carrier frequency and different signal bandwidth conditions show that,the fundamental output of the shortwave power amplifier and the NMSE between the output fundamental signal and the original input signal can reach below-47 dB.The second and the third harmonics can be cancelled to below-68 dBc.