1X De-embedding Method Based on Artificial Neural Network

In this paper, we proposed a new 1X de-embedding method based on artificial neural network (ANN). Firstly, the S parameter of fixture + device under test (DUT) and the DUT itself are taken as the input and output of the ANN, respectively. Through numerical example, this ANN shows a higher accuracy than the traditional 1X method. Especially the ANN method gives a correct DUT S parameter at the boundary of the frequency band, while the traditional 1X method fails. Furthermore, an optimized ANN is proposed, where the prior information is added. This optimized ANN has a more accurate prediction.