Behavioral modeling & analysis of digital predistorter for Hpa using neural committee machine
Loading...
Date
2021-03
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
G.B. Pant University of Agriculture and Technology, Pantnagar - 263145 (Uttarakhand)
Abstract
For efficient use of HPA, HPA should be operated in high-power region near saturation, which severely distorts the information signal. Due to non-linearity, there always exists an elemental tradeoff between the linearity and the efficacy of the HPA. These nonlinear distortions become more vulnerable when communication system demands high data rate transmission, achieved using bandwidth-efficient modulation techniques, for instance, QAM or multicarrier signals in OFDM having large envelope fluctuation with high PAPR. For improving the linearity-efficiency trade-off along with spectral efficiency, an efficient DPD technique must be required. The distortion effects of PA are more critical in the presence of other transmitter impairments i.e. I/Q imbalance and DC offset which further degrade the performance of communication systems. Since conventional DPD does not mitigate these effects in a single step, hence in this work an attempt to formulate a single step solution to solve this problem is made. In this thesis, the NN committee machine i.e. MoE is studied to implement behavioral modeling, predistorter, and DPD for linearizing the DUT. Firstly, MoE is used to implement the behavioral model of the memoryless Saleh model and then DPD characterization is performed for it, and the simulation results present that MoE performs both the tasks in an efficient manner. Then, MoE is used to perform behavioral modeling of dynamic PAs/transmitters in the presence of other transmitter impairments and is also used to implement predistorter for it. Simulated results show that MoE very closely follows the characteristic of dynamic PAs/transmitter and its inverse characteristic in the presence of other transmitter impairments. After that augmented method and kernel method are incorporated with MoE for enhancing the performance of solo MoE. The simulated results for that show the improved performance with respect to its solo performance. At last, measurement is conducted for validation of digital predistortion and the obtained results show that spectral regrowth in the adjacent band in the system is substantially reduced with respect to no DPD case.