swoshie:
each 10ms decide wich input is the best (most pulses).
I don't understand your criteria for 'best', but assuming that pulse counting gives you what you want then doing that on eight inputs seems simple enough. I don't know what's involved in switching your multiplexer but it sounds as if you already know that part is possible. The only thing I'd change in your approach is to poll the inputs and count state transitions directly rather than try to use pulseIn to measure the pulse lengths.