The "too hard" part of the exercise would probably be building the necessary discrete receiver - everything these days is a "somewhat" software-defined radio with no analog signal path - RF in, digital out, and generally little opportunity to open it into a wideband mode. A lecturer of mine last semester (?earlier?) was genuinely shocked when I talked about an ASK transmitter based on discrete transistors - "everything is in a chip now, and people just build the application note" :)
Ah, you're assuming that the PA [if any] input terminal is the only route for signal to get out of the box? I do wish I could design equipment to your assumptions :) Ferrite and mu-metal futures are looking good...