I'm working in an application similar to the one described in the paper attached. Does someone have experience using Walsh Functions for RMS calculation? I think they are not used much in the industry perhaps because they are not so practical (?). I'm using a STM32 microcontroller with no FPU and I need to find RMS of a 50/60 Hz signal like in the paper. Is this a good approach to try or are there better ways?

 

How about just taking many samples of the waveform (sample rate high enough to catch the highest frequency of interest in the non-sinusoidal waveform) and then calculate the RMS value from those (square root of the sum of the squares)?