Actually this is something that I have been thinking about, it is far from impossible

An "one-click-safe-from-NSA-voicecalls" solution;

• 1.) prerequisities; Both A&B subscribers have the encryption software installed in their devices
• 2.) normal CS/PS voice call initiation
• 3.) when both parties have verified that the other end is who it is supposed to be, they enter secure mode by starting the encryption application
• 4.) the encrypting applications take over the voice channel, users are cut out from audio;
  - Layer 1 is audio modulation with fairly low bitrate and similar characteristics as speech range so that it passes reasonably unaltered through echo cancellation and other mangling that RAN does to it.
  - TCP/IP over that carries the connection data
  - SSL handshaking takes care of protecting the connection
  - finally the voice connection is now run on top of the secure connection
• 5.) users have end-to-end encrypted voice channel for the duration of the rest of the call

There was a study and demonstration set up with N900 devices, I belive. It would be easy to have this kind of system on multiple platforms, you'd not be limited to use this only on Jolla-to-Jolla calls.


The scenario I suggested above will of course not hide your device location, nothing can be used to do that if you want ot be on a public cellullar network, but there are ways of hiding who you are communicating with;
Imagine that instead of having a direct voicecall between A&B subscribers you could also set up the system so that both parties have their own connection point in their own controlled networks. After each party sets up connection to their own systems, call could be routed via TOR or similar approach between the connection points