Pattonias said:
We were able to intermittently pick up cell phone calls in our radio class in the army. I would imagine it wouldn't be a stretch to listen in on a call if you built a radio with that purpose in mind.
Not even a little bit of stretch. Agilent (and others like Rhode-Schwartz) makes the
http://www.home.agilent.com/agilent...536881883.00&lc=eng&cc=GB&pselect=SR.Looking" that exists.
With these, you can view the frequency domain of the comm band (GSM, WiMax, etc.) and pick off any transmission and then dive into any cell site code and pull out the audio data stream. Data streams can also be picked off. All passively. And even with cell hand-offs handled automatically (as long as the signal is strong enough from each cell).
It's quite trivial and "cheaply available" for $50K-$150K per rig. The NSA and other countries' intel/security services are their biggest customers for these (not just the USA!).
The only real issue: processing the audio to locate relevant conversations, translate them, recognize the value and then route along with billions of other "suspicious traffic". A human has to be at several parts of the loop with each being another weak point.
That's really the limitation - especially if the conversation is in Arabic, Pashtun, or any other similar language. There aren't enough people on the planet who can 1) speak the native language, and 2) speak English well enough, and 3) interpret the relevance of the conversation within military strategic/tactical utility, and 4) are willing to sign up for the US agenda and allegiance. The co-dependent probability ends up being <1/4E9 pretty easily.
There was recently a call from DARPA to create dictionaries of idiomatic expressions and cultural metaphors in all the world's languages. Why would DARPA care about that? Because if you are doing machine translation and machine triage on gathered ELINT like this, you need that for when folks "talk in code" using common linguistic metaphors of their language (intentionally or not).