I’m just getting to watch this now. I am seeing repeated references to a Private key embedded into a hardware radio. I have serious misgivings about this as I am aware of extensive hardware attacks upon ‘Secret’ hardware designs  and I think this is going to end up being a type of security through obscurity.
I am going to challenge this idea and ask the team to consider the repercussions of pre-storing a private key in hardware:
- The private key is known to more than one entity.
- The private key is by stated design, unchangeable and thus this will make hardware that is not upgradable if a future vulnerability is found.
I also am concerned this could be a step away from using COTS commodity hardware for the radio component.
That being said, I am interested in SDR implementations that are software at the core and extremely upgradable by implementing patches and feature upgrades.
 extracting the private key from a trezor
Side channel attacks are not as difficult as many people think. A simple power analysis requires only a simple oscilloscope and that can hardly be called expensive laboratory equipment. You also need basic soldering skills and deep knowledge about the code that is running. It took only a single recording of the computation of the public key, to recover the private key. On the bright side, this simple side channel attack can be mitigated by using constant-time code and as I showed this code does not have to be slow.