Running the modules at full speed (16MHz) and adding the current consumption of the radio blocks on the TX and RX modules I noticed that the receiving will cause a lot of battery drain:

RX: ~13mA (radio) + ~9mA CPU = ~22mA

TX: ~11mA (radio) + ~9mA CPU = ~20mA

So I've dropped the CPU frequency to 1MHz and "PWMed" the radio block disabling it for a period of time (around 150ms) and enabling it for a short period (~50ms). This will enable a power saving around 75% on the radio alone. The CPU will drop to around 1mA making the total current of :

RX: ~4mA (radio) + ~1mA CPU = ~5mA

This is more acceptable.

The 12V rail current drain will be even lower as the switch mode power supply will draw less than that from the 12V.

I'm going to make some real measurements and post them here.

On the TX side I've also applied this power saving scheme but I had to apply less "off time" to avoid having to press the button a long time for the receiver to acknowledge the signal. It's about 33% off time. but I'll also try to lower the TX power so I can get aditional savings.

TX: ~8mA (radio) + ~1mA CPU = ~9mA

A CR2032 battery has around 240mAh of capacity so this results in 26hours of remote module autonomy.

Assuming a average button press of around 3 seconds (will be lower than that but still) this gives an autonomy of 31200 button presses.

No problem here, I gess :)