During the second half of every year we start to collect data to get a view of what has occurred in the last year and what we anticipate for the next year. Over the last two years we have noticed an industry trend when surveying the marketplace—an increase in the implementation of non-standard architectures. This trend is not restricted to the non-military embedded market segment (commercial, industrial, automotive, and so on); quite the contrary, it seems to be of extreme interest to military designers.

I guess the first thing that needs to be done is to describe exactly what we mean by the term “non-standard architectures”. For the purpose of this column it’s any electrical assembly that is more than one chip, but less than a complete system—something that is part of a larger electronic system. In addition, it would not be designed and built, for example, to any open standard architecture or bus structure such as PC/104, VME or CompactPCI.

The current increasing appeal for these electronic elements can be cost, packaging size, specific and dedicated functionality, power savings, low hardware and software design overhead, and other metrics. In previous years, utilizing an industry standard bus architecture was the preferred solution because it provided some defense against the vendor or its product being discontinued. Although there were rarely, if ever, any direct replacements if a board was discontinued, having a variety of similar products based on a standard architecture made the pain and cost of replacing a board much more manageable. That being the case—what’s happened recently that encourages companies to venture away from open systems based on standard architectures?

One answer in the non-military market may be that with the increased speed at which silicon becomes obsolete there is minimal added risk in having a board or subassembly going obsolete. Instead of using bus-based boards a designer can develop a major electronics subassembly on a large board or multiple boards utilizing many small embedded proprietary components, put some metal around it and exchange data with other major subassemblies using some high-speed comms or serial bus. By the time any of the embedded components on these subassemblies become unavailable the company is already starting on their next generation end product. A concept like this eliminates the cost of the bus overhead, and you can make your end system any size or shape desired. The biggest problem is finding, evaluating and designing in each of these small embedded components to get what you need.

These end systems tend to mirror the sales philosophy of the PC market. You introduce a product, market the heck out of it for a short period and produce just enough to get you to the point where the replacement product comes on line. Suppliers of a system developed like this presumably can end up in a pretty good position. Unlike the PC world where virtually every system runs on a Microsoft program, these systems can run on software that is fairly unique—making it more advantageous for customers to keep coming back to the same supplier for new or upgraded systems rather than finding another source.

If obsolescence and technology insertion are as big a concern to the military as we are being told, then why is there an increasing migration to a design philosophy that adds yet another layer of obsolescence issues? With military programs having a life span of 10 – 20 years, and some even longer, having not only components go obsolete but now proprietary embedded electronic assemblies going obsolete seems to compound the problem. At least with components there are people who make a very good business in supplying obsolete parts or work arounds. How many companies will be willing to provide the same service for the Schmedlap Winken and Blinken module?

At the highest level some of this may just be that the primes are trying to carve out more of the available defense budget for themselves and trying to secure future contracts (See COTS Journal, Publisher’s Notebook, April, 2004). But, the concept of having a captive customer has not been missed by the suppliers of these non-standard architecture assemblies. Unless the DoD finds a faster and easier way to field programs and better facilitate technology insertion into its programs, drifting away from open architectures and standards philosophy may offer low cost solutions now but it will be disastrous in the very near future. The House Armed Services Committee’s Subcommittee on Tactical Air and Land Forces is looking into the benefits of small companies providing state-of-the-art technology products to the military. I hope that during this process they have the foresight to see the benefit of not only COTS solutions, but also the benefit of standards and open architectures.