The key to making a system like the one proposed in that paper work is very precise matching of the frequency and phase of the reference signal to the current signal.
How will you determine the frequency of the current? What is the resolution and accuracy of that measurement? Can you generate the reference signal with resolution and accuracy equal to or better than that of the current measurement? The ultimate accuracy limits of your system will depend on the frequency and phase matching as well as the amplitude measurements.
You do not need simultaneous sampling to do these things, although it is much easier if you have that. With a multiplexed A/D converter you need to account for the delay between samples on different channels.
Are you familiar with phase-locked-loop and lock-in amplifier techniques?