Via this Wired piece, I see that the US Special Forces are giving up on the various "Land Warrior" projects to load down soldiers with specialised electronic gear - some readers may remember the BBC documentary on one of MOD's efforts in this line and the image of a file of soldiers attempting to move stealthily across Salisbury Plain with enormous objects like big plastic mushrooms lashed on top of their helmets. Instead, they've issued an RFP for software applications running on Android-based devices to achieve the same aims.
The RFP is here, and what an RFP it is too. It is, among other things, clear, and clearly drafted by someone with substantial technical competence. You try finding many contract managers who know what RFC 5740 specifies. From a technical point of view it's pretty demanding: multiple video streams, reliable delivery, in an environment of restricted connectivity - rather you than me. (This 2008 RFP may be part of the explanation.)
However, it's also true that a valid strategy for delivering high bandwidth traffic like video is to shift it from the classic unicast (i.e. one stream per user, from the same source) to a broadcast or multicast route. I wonder if whoever answers it will be an early user of the Stream Control Transmission Protocol, which among other things allows multi-homing at the connection level, so that the same connection between two logical addresses can involve more than one physical source?
This also reminds me of the mid-2000s IP Multimedia Subsystem hype - collaborative whiteboarding was an example use case that came up in literally every vendor presentation, and they would occasionally do demonstrations at conferences, which always turned out to be really awful. Sometimes this was because the server was back in Finland and the endpoints were roaming on a Singaporean operator's 3G network, with hilarious latency consequences. Sometimes it was because IMS just wasn't a very good idea.