A new methodology for in situ calibration of a neural network-based software sensor for S-parameter prediction in six-port reflectometers

In this work, a neural network-based software sensor is proposed for determining the reflection coefficient from measurements obtained by a six-port reflectometer. The proposed software sensor is able to cope with the nonlinearities and noise inherent to the measurement electronics, without needing additional calibration. To extract data for the calibration, a new method that allows in situ calibration is applied. Experimental evidence of the feasibility of the proposed method is given using a simulation testbench.