How does the predator network prevent outside control by an attacker?

Question

I read several articles saying some of the command and control fleet for the US robotic attack planes may have been infected with a virus.

Does this command and control system communicate with the planes using the internet on a giant VPN of sorts? What systems are in place to prevent these robot planes from becoming controlled by someone else?

Please limit discussion to unclassified info to keep everyone out of trouble.

Answer 1

See: wiki on SIPRNet and NIPRNet.

Essentially the two are airgapped, and should not be interconnected at any point. In fact, measures are taken to prevent that very thing. Nevertheless humans are on both networks, which means the results are mixed.

Answer 2

In fall 2011 Windows computers used load and store data associated with unmanned ariel vehicles at Indian Springs Air Force Base were infected with key loggers.

It appears to me (former USAF navigator) that the infected computers were used for mission planning, not for directly controlling the aircraft. It's very possible that these computers could access the civilian internet. In any case, Stuxnet showed how to extract captured data from an air gapped facility. The Air Force has to assume that data entered in to the infected computers has been compromised.

The other known security fiasco associated with the Air Force UAV program was the fact that the UAV video downlinks were for a long time unencrypted and could be monitored with civilian TV satellite receivers. Rumor has it this vulnerably was discovered during interviews with captured Iraqi insurgents.