April 02, 2008
With a recession on the way, or already here, some technologies stand a better chance of receiving funding or being purchased. Technology that can directly make a connection to an end user benefit is the secret here. With 2008 S&T funding reduced by an astounding 20.1 percent to $10.9 billion, the military science fair is over, kids, as only DARPA gets to play with the fun toys. Instead, the DoD prefers to buy only ?right now? capabilities.
At the top of the list is anything that helps the U.S. prosecute the Global War On Terror (GWOT) while saving the lives of service personnel. Communications gear, crypto and secure software are also hot, as long as an end-user direct benefit can be made. Low power or devices that extend battery power are crucial. And any hardware or software platform that can build off of a legacy system's software or hardware puts you several steps ahead of the game and your competition. That edge might just get you funding.
One in a crop of interesting new doodads is Ambric's Massively Parallel Processor Array, being targeted at defense and security applications. It's never a good time to be a processor company - much less start one - but these folks have been at it since 2003 and have spent only $21 million of Series A funding. Their tiny burn rate has given them big credibility with the DoD, as has the CPU's amazing performance. Designed as a stream processor at the front end behind a sensor, the 1 TeraOPS, 336-processor device maximizes MIPS/mW: Think of it as a Cray-on-a-chip. That means a lot less power in a much smaller space, putting to shame general purpose DSPs (5-20X over a TI C641x, they say), or FPGAs that require complex RTL coding.
As we went to press, Ambric was literally en route to present a paper about on-the-fly partial CPU reconfiguration - a particularly handy trick for data mining, cognitive radio, and myriad other military algorithms. DSPs can't do that, and Xilinx's FPGAs are only now sort-of talking about partial reconfigurability.
One of the ways Ambric might find its way quickly onto the battlefield is deployed in a VITA VPX chassis mounted in an MRAP for counter IED operations (more on IEDs in a moment). Itself a shining example of rapidly developed and deployed technology, the Mine Resistant Ambush Protected armored vehicles are being rushed to Iraq and Afghanistan to replace the woefully inadequate HWMMV. Some of those Hummers might also be finally equipped with VITA's now-available VPX technology, which is all about the I/O - lots of I/O.
Far as I can tell, Curtiss-Wright Controls Embedded Computing recently became the industry's very first COTS vendor to announce a VPX design win - for the Marines' Ground/Air Task Oriented Radar (G/ATOR). It's not in an MRAP, it's in a Hummer. And it is an FPGA-based DSP board integral to the Active Electronically Scanned Array (AESA) radar used for tracking air- and ground-based targets, as well as general air traffic control. Here too, CWCEC had the technology ready to go, and VPX was the ideal form factor for quick field deployment. The prime, Northrop-Grumman, awarded CWCEC $4.3 million for their efforts. Kudos.
And with technology budgets thinner now than at any time during the Bush Administration (O&M for two fronts is taking the lion's share), extending the life of anything - particularly software - is key. No one has the time or money to write code from scratch if modules can be reused. I call this Legacy Software Migration, and it will form the basis of a new, regular column in this magazine starting in May. But in this issue, contained in our "RTOS updates and trends" feature on page 16, the most common theme I found from our industry survey is how easy it really is to migrate forward software and hardware.
It's all about reuse, folks. For instance, if you believe Ada is so 1990, then either Aonix or DDC-I can move you into safety-critical Java. Or MapuSoft can "easily" migrate you from one hardware platform to another while preserving legacy code. As well, Impulse Accelerated Technologies - from which we'll hear in a future issue - has a slick way of abstracting an obsolete PowerPC chip into an FPGA. Besides that, Quantum3D has a software-based GPU called IGL178 that is obsolescence-resistant. Common theme with all of these: saving time and money and getting a useable system to the field, double-time.
And no one knows more about double-time than the Joint IED Defeat Organization (JIEDDO). These folks are all about rapidly deployed tech rushed to the front lines to counter IEDs and save lives. At the well-attended recent Military and Aerospace Electronics Forum, JIEDDO Deputy Director Dr. Robin Keesee talked about whittling 1,335 proposals down to 17 major funded initiatives in an astounding 68 days. Technology to save lives can be fielded, he said - proposal to deployment - in a mere six months! Compare that to JTRS, which is going on maybe eight years.
These represent the kinds of technologies that are getting funding as a recession looms. Not every VPX board, FPGA SDR, or C cross-compiler will see such a short time-to-money. But making sure your technology or your program has a direct end-user benefit will dramatically enhance the chance of funding or program success. Don't let DARPA have all the fun.
Chris A. Ciufo