I don't understand why coverage measurements should be done at SF12 as a "normal" node sends at SF7? so we would then artificially extend the coverage...
at SF12/ 125khz / CR4/5 sending just a 16 bit integer like kersing does, already costs about 500 ms airtime (at SF7 20ms), and 19 bytes (like I do) about 1000 ms (at SF7 only 40ms)
btw. I'm also using an accellerometer to limit reaching the duty cycle too quickly (and max. 30s airtime per day) 