[[update: be sure to look at the comments here for some sound defense of SCA.]]
Dr. Ronald Jost spoke this morning at the second day of the IDGA’s Software Radio Summit. The Deputy Assistant Secretary of Defense for C3 Space and Spectrum said essentially that there was a major disconnect between what industry saw as the advantage of software defined radios — the programmable radios capable of being configured for multiple types of communications and being upgraded via software– and what DOD wanted them for. While industry is dazzled by the potential to program all varieties of new custom waveforms, the DOD, he said, just wanted to use SDRs to help consolidate its communications networks toward a single, IP-based topology–and save some money on maintenance of the equipment.
But it isn’t happening the way the DOD wanted it to because of a number of reasons: first of all, they’ve ended up buying every possible configuration of SDR possible hardware under the JTRS program and other procurements, and so the advantages of a simple operating environment have thus far been lost. “You’re killing us,” Jost said to the vendors. “We can’t port waveforms from one radio to another in 1 or 2 months with 2 or 3 people,” which was the goal. Instead, he says, “it takes 100, 200 people a year.”
The vendors aren’t the only culprits, obviously. The real problem has been that the requirements were fouled from the beginning, and didn’t really spell out what DOD hoped to get out of SDRs — migration from a universe of different radios with unique transmission profiles and protocols down to a smaller set of waveforms and IP-based protocols; and modular design that allowed for low cost (5 to 10% of initial cost) upgrades to extend the life of the equipment out into the future.
The Software Communications Architecture was supposed to make everything easy to port around for JTRS radios– it’s based on CORBA, POSIX and some other very old standards. But it didn’t, because, first of all, the program ran into a giant pile of scope creep early on, and it’s now nearly 20 years into it and has only delivered one interim set of radios to warfighters. Also, early on, the programs under JTRS were all being run by the various services without joint coordination, and…well, the end result was that the platforms were highly divergent.
To say the least, vendors have not been excited about the government’s licensing terms for JTRS — that the government owns all the code for perpetuity. JTRS maintains a library that radio suppliers can check out code from to port to their radios–theoretically reducing the cost of entry. And multiple suppliers are supposed to be certified for each platform set, creating competition. But there’s still some considerable barriers to entry for would-be suppliers (particularly, say, multinational companies and smaller manufacturers).
All of that makes the JTRS platforms pretty expensive. And Jost says that if SDRs based on SCA aren’t cost effective, the DOD will probably turn away from them. DARPA apparently hates SCA (probably because it’s based on CORBA), and the Wireless Network After Next (WNAN) project that BBN is doing with DARPA is software based but not based on SCA, because of the added development cost.
Jost said that SCA doesn’t make that much sense for users on the edge of the network–that would be, handheld radio users–because of the cost. This may come as a surprise to the folks working on JTRS’s Handheld-Manpack-Small Formfit (HMS) radios, and to the JPEO JTRS team which is hoping the services will pick up its Rifleman Radio and HMS-based systems for those very edge users.
There was also the comment by Jost that I Twittered earlier: “No one would put a product on the shelves at Best Buy that required a T-1000, or like a 10Gigabit connection to work properly. They’d go bankrupt. But DOD does that all the time.” He was speaking in reference to some of the new C4ISR applications being pushed out of various DOD acquisition programs. Applications, as he noted, move faster in technology than the DOD’s network does–given it’s going to take 30 years from the startpoint of JTRS to deploy the last GMR radio, there’s a lot of legacy out there, and the term “disadvantaged user” covers just about everyone in DOD who isn’t bolted onto a wired network. So the those creating all sorts of new tools to push data to the warfighter have to be aware of the realities of the DOD GIG — the edges are slow, and they aren’t getting faster quickly.