Linearization of high-frequency Doherty power amplifiers using artificial neural networks

The Doherty power amplifier is an RF design of a class B amplifier configuration that achieves high efficiency by having two amplifier sections. One amplifier section is suitable for lower-amplitude signal conditions. Then a second amplifier is introduced to meet the requirements of higher-level signals without compression. In this way, the Doherty amplifier can provide both linearity and efficiency. The concept of the Doherty power amplifier configuration was invented by William H. Doherty of Bell Telephone Laboratories in ۱۹۳۶, and various studies have been conducted in this field. One of the current discussions in the scientific community is the linearization problem, which in this thesis was addressed by linearizing the Doherty amplifier using a neural network. For this purpose, coding was done in MATLAB software and the M-File environment. And the results obtained by the neural network with the linearization method were compared with the polynomial method and real values. For this, a bandwidth of ۲ GHz was considered. The results showed that the trained network can perform up to ۹۸% linearly.

کلیدواژه ها:

Doherty amplifier ، linearization ، high frequency ، neural network