Use of Artificial Intelligence in Classification and Monitoring of VHF Signals in a Software Based Instrumentation System

Abstract

A software based instrumentation system was designed to measure the transient frequency response for a 50 MHz signal with a precision better than 0.3 ppm. Long short-term memory (LSTM), an artificial recurrent neural network (RNN) architecture was used to detect and classify features on signals generated by this system. Dropouts in signal were detected and characterized with an accuracy better than 78%. The concept of software based instrumentation was implemented using a PXI based instrumentation system. The software solution was implemented in LabVIEW, Matlab and LabWindows/CVI.