Independent Component Analysis for Signal Crosstalk Elimination in Infrared Interferometry
Infrared interferometers are optical devices that can measure optical path-length differences by measuring changes in the refractive index. Several arrangements can be deployed, from single channel devices to multichannel double color heterodyne interferometers. These type of devices are typically used to recover the spatial electron density profile of fusion plasmas. This process involves measuring precisely the phase differences of the intermediate frequencies. A source of error that affects the measurements is the crosstalk that appears due to the coupling of the signals into the different channels of the interferometers. The inter-channel coupling of the signals is extremely difficult to eliminate specially when the probing frequencies are close to each other. In this paper it is shown that Independent Component Analysis effectively eliminates inter-crosstalk coupling in such devices. Furthermore, it is shown how the Signal-to-Noise ratio is dramatically increased when this technique is used.