The Study Of PA Macro-model Behavior Base On Fuzzy Logic

RF power amplifier has important application in the emission system of wireless communication, and is the most important apparatus of base station transmitter. The design and research of RF power amplifier has always been an important subject during the development of wireless communication. The establishment of RF power amplifier model to replace the actual hardware circuit model, and the conduct of systematic assessment has a great significance on the shortening of development time and the saving of research cost. Meanwhile, with the development of integrated circuit, the circuit size increased and circuit complexity improved, thus resulting in higher requirement for modeling. Therefore, the study of modeling technic and amplifier model has profound meaning.This paper focuses on the establishment of RF amplifier behavioral model, and the application of the integrated approach of fuzzy logic and neural network, as we called adaptive Neuro-Fuzzy inference system (ANFIS), in the modeling of RF power amplifier. The modeling technology of combining fuzzy logic with neural network is an important subject in current modeling technology development. The traditional amplifier behavior model can be divided into two types, with memory and without memory. This paper will study and compare the traditional behavior model, focusing on the polynomial model with memory and several types of neutral network model.This paper will study the usage of ADS software to design amplifier circuit, extraction of required data of modeling through simulation, usage of fuzzy netural network model for modeling, and the test of established amplifier model. Then we use this to compare to the simulation result of the circuit, including transient steady-state waveform, the spectrum of different periods, power compression characteristic curve, power gain characteristic curve and so on. We can see from the result of comparison that the accuracy of the established model is relatively high. This modeling method is of great significance, because we can apply this modeling method into different circuit modeling to improve the efficiency of modeling. And how to continuously improve the efficiency of modeling is also an important research field of and the development trend of modelling technology.