Here we are at 2010. Here's wishing you a happy and prosperous year ahead. While we may not have flying cars and manned spaceships exploring Saturn's moons quite yet, 2010 certainly feels like the future. Military electronic systems-across all branches, manned and unmanned, large and small-continue to head in the direction of increasing complexity as programs require ever more functionality, autonomy and intelligence.
That's all good for our segment of the industry because, naturally, that drives demand for faster embedded computing and other more sophisticated electronic subsystems of every kind. That upward trend in complexity, while perfectly in line with the idea of providing always improving the warfighter's ability to do their jobs better,does come at a price. It's no big secret that the defense industry has felt the downsides of growing complexity in a more profound way than other industries. Costs of major programs inevitably grow significantly over budget while average program development cycles have grown by up to 20 percent over the last decade-and predicting accurate development cycles seems to be getting harder.
All kinds of attempts have been made to mitigate those overruns through the "top down" acquisition. Some progress in terms of understanding the weaknesses in structure, process and governance of acquisitions has been made, but significant cost/schedule overruns still persist. The latest efforts at tackling these issues come from the Defense Advanced Research Projects Agency (DARPA), taking what could be called a "bottom up" strategy of reforming the engineering process side of things.
DARPA sometimes chooses its research by doing what it calls a Broad Agency Announcement (BAA) process. Last fall DARPA released a BAA soliciting research proposals for a program called META. The goal of META is to "substantially improve the design, integration/manufacturing and verification of complex cyber-physical systems, and particularly aerospace and defense systems such as aircraft, rotorcraft and ground vehicles." The term "cyber-physical systems," as you might guess, is cringe-worthy jargon for people like me in the wordsmithing business. DARPA defines cyber-physical as systems that derive significant portions of their functionality from both software and electromechanical systems. Virtually all defense platforms-aircraft, spacecraft, naval vessels, ground vehicles-and systems-of-systems fall under that rubric, as do automobiles, power grids, air traffic control systems and integrated circuits. I prefer to cut to the chase and use the term "embedded system" instead, but that's just me.
Last month DARPA held an Industry Day event to review proposals from the various companies interested in pursuing the program. Looking at some of the material from the introductory part of the event, I saw a chart from Paul Eremenk's presentation that blew me away. Eremenk is with DARPA's Tactical Technology Office. It should be a fascinating comparison between the mil/aero industry, the automotive industry and the integrated circuit industry. The chart graphed complexity-measured in parts count plus lines of software source code-versus design, integration and testing time (in months). The IC industry, as the chart showed, stayed at relatively fixed development cycle time as complexity increased. The automotive industry meanwhile has moved steadily down in development time as cars have grown in complexity. Meanwhile aerospace and defense systems have, over the same timeframe, ramped ever upward in development time and associated cost as complexity increased. To me, the chart is a very vivid illustration of what happens when some industries revamp their design flows, and others don't.
As the chart showed, the idea behind the META program is that, while the complexity of aerospace and defense systems has grown considerably over the past half-century, the systems engineering approach-or, more specifically, the design, integration/ manufacturing and test flow-really hasn't changed since back in the days of Atlas missile development and Apollo programs-which were codified in MIL-STD-499.3.
DARPA called for proposed research under the META program to investigate innovative approaches that enable revolutionary advances in this area. Specifically excluded is research that primarily results in evolutionary improvements to the existing state of practice. The ultimate goal of the META program is to make a dramatic improvement on the existing systems engineering, integration and testing process for defense systems. According to DARPA, META isn't intended to revolve around any one particular alternative approach, metric, technique, or tool. Rather it aims to develop model-based design methods for embedded systems-or cyber-physical systems as they call it-that are far more complex and heterogeneous than those common today.
As someone who has closely followed all the major vendors of system modeling tools and frameworks over the years, I'm having a hard time understanding what exactly the META program will result in that's fundamentally different than what's already available in today's very sophisticated embedded system development tools market. That said, there's never been any such effort-to my knowledge-that's been built from the ground up specifically for the defense and aerospace industry. One challenge is the priority that military systems have on rugged environment features and strict goals for size/weight/power minimization at the subsystem and component level. It's all very interesting at any rate, so I'll be keeping an eye on META as it moves forward.
