The ping is 10ms long, thus the ENERGY per ping is 0.75*0.01 = 0.0075joules or 7.5mJ. ... At a rate of 1 per second, gives an effective average power requirement (ignoring circuitry – which should be pretty simple for a dumb pinger) of 7.5mW.
So, why not maybe change the ULB specs to slow the pinging down after the first week by halving the rate for each succeeding week.